Aberrant p53 fuels macrophage immunoregulatory polarization and refractory clinical outcome in urothelial carcinoma

Abstract

Background

Urothelial carcinoma (UC) exhibits extensive infiltration of tumor-associated macrophages (TAMs). As the most frequently altered gene in UC, TP53 is recognized not only as a guardian of genomic integrity, but also as a potential architect of the tumor immune microenvironment (TIME). However, how aberrant p53 status, comprising TP53 mutation (TP53^Mut) and p53 abnormal expression (p53^Abn), correlates with TAM phenotypic and functional heterogeneity and their reciprocal interactions within the TIME landscape remains incompletely defined.

Methods

This study integrated single-cell RNA sequencing, immunohistochemistry, and flow cytometry to delineate TAM phenotypes and map distinct immune landscapes stratified by p53 status. Clinical relevance was assessed in a multicohort dataset of 962 patients with UC with comprehensive clinicopathological annotations.

Results

Single-cell profiling revealed distinct TAM subsets closely linked to p53 status. Interferon-primed TAMs, marked by CXCL9, CXCL10, CXCL11 and IDO1, predominated in p53 wild-type (p53^WT) tumors, whereas immune regulatory TAMs (Reg-TAMs), marked by SPP1, MARCO, SLC2A1 and INHBA, were enriched in p53^Abn tumors. This divergent distribution highlighted a profound, p53-associated TAM heterogeneity. Clinically, high TAMs infiltration was consistently associated with worse overall survival predominantly in p53^Abn or TP53^Mutpatients across three independent cohorts (Zhongshan Hospital, Fudan University Shanghai Cancer Center and The Cancer Genome Atlas). Furthermore, high TAMs infiltration correlated with poorer responses to both platinum-based chemotherapy and anti-programmed cell death-ligand 1 immunotherapy specifically in the p53^Abn or TP53^Mut subgroup. Notably, aberrant p53 status was strongly coupled with a distinct metabolic reprogramming in TAMs, favoring lipid catabolism and immunoregulatory polarization. Trajectory analysis revealed a skewed differentiation axis in p53^Abn tumors toward terminally differentiated Reg-TAMs. Immunohistochemistry and flow cytometry validated increased immunosuppressive TAM subsets in p53^Abn UC. Additionally, p53^Abn tumors exhibited an altered CD4⁺ T-cell landscape, enriched for suppressive subsets (regulatory, exhausted, and metallothionein-expressing T cells) coexisting with Reg-TAMs, potentially associated with immune suppression via the predicted SPP1-CD44 axis.

Conclusions

In UC, aberrant p53 status correlates with a remodeled TIME, characterized by metabolically reprogrammed, immunosuppressive TAMs. These TAMs co-evolve with suppressive CD4⁺ T cells, suggesting a coordinated immunosuppressive network potentially involving SPP1-CD44 signaling. Together, this specific genotype-phenotype coupling delineates a profoundly immunosuppressive ecosystem that likely impedes immune clearance and limits therapeutic efficacy.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• While urothelial carcinoma (UC) is characterized by substantial tumor-associated macrophages (TAMs) infiltration and frequent aberrant p53 status, the specific association between p53 status and the TAM functional heterogeneity of the immune microenvironment remains poorly characterized.

WHAT THIS STUDY ADDS

• This study identifies a robust link between aberrant p53 status and an immunosuppressive microenvironment, defined by the accumulation of metabolically reprogrammed TAMs that co-evolve with suppressive CD4⁺ T cells, potentially interacting via the predicted SPP1-CD44 axis.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings provide valuable insights into the association between aberrant p53 status and an immunosuppressive tumor immune microenvironment, highlighting potential therapeutic strategies targeting TAMs and the SPP1-CD44 axis. By identifying biomarkers associated with immune resistance, this research may help guide the development of more tailored treatment strategies for patients with UC with aberrant p53, potentially improving patient outcomes.

Introduction

Urothelial carcinoma (UC) is a common and aggressive malignancy with persistently poor long-term survival.¹ Platinum-based chemotherapy has historically defined first-line care, but the advent of immune checkpoint inhibitors and targeted agents, including FGFR3 inhibitors and nectin-4 or HER2 antibody-drug conjugates, has diversified treatment options.² Yet, most patients ultimately develop resistance, driven by a combination of tumor genomic alterations and the heterogeneity of the tumor immune microenvironment (TIME).

Among genomic drivers, TP53 is the most frequently mutated gene in UC and a key guardian of genomic stability.³ Its inactivation accelerates genomic evolution,⁴ promotes features such as extrachromosomal circular DNA formation, and fosters immune-evasive phenotypes.^{5 6} Our previous work has shown that variations in TP53 mutation profiles or protein expression levels correlate with differential responses to chemotherapy and chemo-immunotherapy, likely reflecting underlying genomic and immune heterogeneity.⁷ While p53 abnormalities are known to dampen innate immune signaling,⁸ its broader impact on shaping immune cell composition and function within the TIME remains unclear.

Among immune components of the TIME, tumor-associated macrophages (TAMs) are key mediators of UC progression and treatment response.^{9 10} These cells adopt diverse phenotypes that can either support antitumor immunity or promote tumor progression.11,13 In UC, abundant TAMs infiltration correlates with suboptimal responses to immunotherapy,^{14 15} but the environmental cues shaping TAMs heterogeneity, and the extent to which p53 status is associated with this process, remain poorly understood.

In this study, we systematically characterized TAMs composition and functional states in UC stratified by p53 status and examined their associations with patient prognosis and treatment responsiveness. By investigating the specific metabolic reprogramming and phenotypic shifts of TAMs in the context of aberrant p53 status, we aim to uncover potential mechanisms of immune evasion and therapeutic resistance. Our findings provide a rationale for biomarker-guided stratification and inform the development of precision immunotherapy strategies in UC.

Materials and methods

Patient cohort

A total of 962 patients with UC from four independent cohorts were included in the survival analysis. The local cohorts consisted of 297 patients from Zhongshan Hospital (ZSHS cohort, n=135) and Fudan University Shanghai Cancer Center (FUSCC cohort, n=162), who underwent radical cystectomy between 2002–2014 and 2008–2012, respectively, with tumor tissue microarrays (TMAs) and detailed clinical annotation (online supplemental tables 1 and 2). Overall survival (OS) was defined as the interval from surgery to death from any cause or last follow-up. The public cohorts included The Cancer Genome Atlas (TCGA) cohort (n=391) and the IMvigor210 cohort (n=274). Detailed inclusion and exclusion criteria for the public datasets were provided in online supplemental methods.

Single-cell RNA sequencing

A total of 23 freshly resected UC specimens (p53 wild-type, p53^WT: n=12; p53 protein abnormal expression, p53^Abn: n=11) were obtained from 23 patients undergoing radical cystectomy at our local clinical centers, following written informed consent, and served as the primary discovery cohort (Shanghai Urothelial Carcinoma Research Group cohort, cohort). Single-cell RNA sequencing (scRNA-seq) libraries were constructed using the Chromium Controller system and Chromium Single Cell 3’ v3 Reagent Kits (10x Genomics, Pleasanton, California, USA). Sequencing data were processed in R environment (V.4.3.3) using the Seurat package (V.5.2.1), with stringent quality control criteria to remove low-quality cells (genes detected per cell <300 and mitochondrial gene percentage >20%). Batch effects were corrected using the Harmony algorithm, followed by cell clustering with the Louvain algorithm implemented in the FindClusters function. For two-dimensional visualization, Uniform Manifold Approximation and Projection (UMAP) was applied. To focus on immune cell populations, cells expressing PTPRC (CD45) were computationally extracted for all downstream analyses. Cluster-specific marker genes were identified using the FindAllMarkers function with the Wilcoxon rank-sum test, applying the following thresholds: (1) log₂ fold change >0.25; (2) adjusted p value<0.05; and (3) min.pct >0.1. An additional five UC specimens (p53^WT: n=2; p53^Abn: n=3) were independently collected from the FUSCC as a validation cohort. For this cohort, CD45⁺ immune cells were prospectively enriched by fluorescence-activated cell sorting (FACS) prior to scRNA-seq library construction, providing orthogonal validation of immune cell findings. Detailed single-cell processing and analysis procedures were described in the online supplemental methods.

Immunohistochemistry staining

Single and double immunohistochemistry (IHC) staining were performed on formalin-fixed, paraffin-embedded TMAs from the ZSHS cohort according to established protocols.¹⁶ Based on p53 expression in tumor cells, cases were categorized into two subgroups: (1) p53^WT (1–80% nuclear staining, variable intensity), and (2) p53^Abn (absence of tumor cell nuclear staining with a positive internal control, or uniform and intense nuclear staining in ≥80% of tumor cell nuclei), in accordance with clinicopathologic guidelines.717,19 IHC scores for immune cells population and functional immunoregulatory molecules were determined as the mean values from two independent pathologists who were blind to clinical data and standardized to represent infiltration levels. High infiltration was defined as the top quartile (highest 25%), and low infiltration as the remaining 75%. Detailed protocols and antibody information were provided in online supplemental methods and online supplemental table 3.

Flow cytometry

To assess the infiltration levels and functional phenotypes of TAMs and CD4⁺ T cells, freshly resected UC tissues (p53^WT: n=37; p53^Abn: n=40) were collected from four local hospitals in Shanghai (ZSHS, Ruijin Hospital, FUSCC, and Shanghai General Hospital), following written informed consent from all participants. Detailed flow cytometry (FCM) procedures, gating strategies and antibody information were provided in online supplemental methods and online supplemental table 4.

Processing of bulk RNA sequencing profiles

For transcriptomic analyses, gene expression levels for each specimen were normalized to fragments per kilobase of transcript per million mapped reads. The infiltration score of TAMs and interaction signature score between clusters of immune cells were calculated using the single-sample Gene Set Enrichment Analysis (GSEA) algorithms based on related genes expression as previously reported (genes listed provided in online supplemental table 5). GSEA was performed in the TCGA cohort to identify differences between TP53 wild-type (TP53^WT) and TP53 mutant (TP53^Mut) UC. Example GSEA plots were generated using the gseaplot2 function.

Statistical analysis

Survival analyses were performed using the Kaplan-Meier method, with differences between groups compared by the log-rank test. Multivariable Cox proportional hazards models were used to estimate adjusted HRs and 95% CIs, incorporating established baseline clinical and molecular covariates. To evaluate whether prognostic associations were modified by genomic background, formal interaction terms were integrated into the Cox models. Pearson’s χ² test or Fisher’s exact test was applied for categorical variables, and the Mann-Whitney U test for continuous variables. All statistical tests were two-sided, and a p value<0.05 was considered statistically significant. Analyses were performed using IBM SPSS Statistics V.26.0 and R V.4.3.3.

Results

Aberrant p53 status characterizes a divergent polarization landscape of tumor-associated macrophages in urothelial carcinoma

To systematically investigate immune ecosystem alterations associated with p53 status in UC, we analyzed scRNA-seq data from our discovery cohort comprising 23 treatment-naïve tumor specimens (p53^WT: n=12; p53^Abn: n=11) (figure 1A). After stringent quality control of the global single-cell transcriptomes, immune cells were computationally extracted based on PTPRC (CD45) expression. This yielded a robust dataset of high-quality immune cells representing major lineages, including CD8⁺ T cells, CD4⁺ T cells, B cells, plasma cells, TAMs, dendritic cells, mast cells, neutrophils, and natural killer cells (figure 1B). These major immune compartments were consistently identified in our independent validation cohort of five prospectively FACS-sorted CD45⁺ UC samples (online supplemental figure 1A). Subclustering of TAMs uncovered six transcriptionally distinct TAM subsets (figure 1C,D and online supplemental methods). A highly concordant TAM subpopulation structure, comprising these six main subsets alongside a minor “Others” group, was successfully recapitulated in the validation cohort (online supplemental figure 1B,C). Notably, interferon-primed TAMs (IFN-TAMs), marked by CXCL9, CXCL10, CXCL11 and IDO1, were enriched in the p53^WT UC, whereas immune regulatory TAMs (Reg-TAMs), marked by SPP1, MARCO, SLC2A1 and INHBA, predominated in the p53^Abn microenvironment (figure 1E). Differential enrichment analysis further revealed this phenotypic divergence: Reg-TAMs exhibited the most profound expansion in p53^Abn tumors (log₂ fold change (FC)=2.08), whereas IFN-TAMs constituted the most prominently depleted subset in this context (log₂FC=−0.68) (figure 1F). This distinct enrichment pattern of IFN-TAMs and Reg-TAMs was consistently validated in the independent validation cohort (online supplemental figure 1D,E). Together, these divergent distribution patterns suggest a notable association between tumor p53 status and the functional polarization of the TAM compartment.

Figure 1. Aberrant p53 status characterizes a divergent polarization landscape of tumor-associated macrophages in urothelial carcinoma. (A) Schematic overview of the experimental design and analytical workflow. (B) UMAP plot showing the major immune cell clusters in patients with UC. (C) UMAP plot showing the six distinct clusters of TAMs. (D) Heat map displaying the scaled expression of the top 100 marker genes for each TAM cluster. (E) Pie charts depicting the proportion of TAM clusters in p53^WT and p53^Abn tumors. (F) Lollipop plot showing the differential enrichment (log₂ fold change) of TAM subsets between p53^Abn and p53^WT tumors. Abn, abnormal; Angio-TAMs, pro-angiogenic TAMs; FCM, flow cytometry; FUSCC, Fudan University Shanghai Cancer Center; IFN-TAM, interferon TAM; IHC, immunohistochemistry; Inflam-TAM, inflammatory cytokine-enriched TAMs; LA-TAM, lipid-associated TAMs; NK, natural killer; Reg-TAM, regulatory TAM; RTM-TAM, resident-tissue macrophage TAM; SUCRG, Shanghai Urothelial Carcinoma Research Group; TAMs, tumor-associated macrophages; TCGA, The Cancer Genome Atlas; UC, urothelial carcinoma; UMAP, Uniform Manifold Approximation and Projection; WT, wild type; ZSHS, Zhongshan Hospital Affiliated to Fudan University.

High tumor-associated macrophages infiltration correlates with poor prognosis in p53-aberrant urothelial carcinoma

We next investigated the prognostic significance of TAMs infiltration stratified by p53 status. In the ZSHS cohort, high TAMs infiltration showed a trend toward worse OS in p53^Abn patients (figure 2A,B, p53^Abn: median OS: 40 vs 90 months, log-rank p=0.090), while no such association was observed in the p53^WT subgroup. Similar patterns were observed in the FUSCC cohort, where the prognostic impact of high TAMs infiltration was more pronounced in p53^Abn cases (figure 2C,D, p53^Abn: median OS: 40 months vs not reached, p=0.027). Consistent findings were also obtained from the TCGA validation cohort, in which high TAMs infiltration was associated with significantly worse OS in TP53^Mut patients, but not in TP53^WT patients (figure 2E,F, TP53^Mut: median OS: 20 vs 97 months, log-rank p=0.003). By incorporating additional clinicopathological factors, multivariate Cox regression analyses showed that the adverse prognostic association of high TAMs infiltration was more evident in p53^Abn or TP53^Mut patients (figure 2G, ZSHS cohort: adjusted HR 1.842, 95% CI 0.958 to 3.541, p=0.067; FUSCC cohort: adjusted HR 3.448, 95% CI 1.686 to 7.050, p=0.001; TCGA cohort: adjusted HR 1.794, 95% CI 1.136 to 2.834, p=0.012). Notably, while the association in the ZSHS cohort showed a consistent directionality, it did not reach conventional statistical significance, potentially reflecting the limited sample size of this subgroup.

Figure 2. High tumor-associated macrophages infiltration correlates with poor prognosis in p53-aberrant urothelial carcinoma. (A–B) Kaplan-Meier curves of OS stratified by TAMs infiltration in patients with UC in (A) the overall ZSHS cohort and (B) the p53^WT and p53^Abn subgroups. (C–D) Kaplan-Meier curves of OS stratified by TAMs infiltration in patients with UC in (C) the overall FUSCC cohort and (D) the p53^WT and p53^Abn subgroups. (E–F) Kaplan-Meier curves of OS stratified by TAMs infiltration in patients with UC in (E) the overall TCGA cohort and (F) the TP53^WT and TP53^Mut subgroups. (G) Multivariate Cox regression analyses for OS stratified by TP53/p53 status in the ZSHS, FUSCC and TCGA cohort. Abn, abnormal; AJCC, American Joint Committee on Cancer; FUSCC, Fudan University Shanghai Cancer Center; mOS, median OS; Mut, mutant; OS, overall survival; TAMs, tumor-associated macrophages; TCGA, The Cancer Genome Atlas; UC, urothelial carcinoma; WT, wild type; ZSHS, Zhongshan Hospital Affiliated to Fudan University.

High tumor-associated macrophages infiltration is associated with chemotherapy-refractory and immunotherapy-refractory outcome in p53-aberrant urothelial carcinoma

We then assessed whether TAMs infiltration influences treatment response in a p53-dependent manner. Among platinum-based chemotherapy recipients in the ZSHS cohort, high TAMs infiltration was associated with worse OS in p53^Abn patients, whereas no such correlation was observed in p53^WT patients (figure 3A,B, p53^Abn: median OS: 39 vs 118 months, log-rank p=0.008). The FUSCC cohort yielded consistent results (figure 3C,D, p53^Abn: median OS: 28 months vs not reached, log-rank p=0.005). In the TCGA cohort, the results also demonstrated a correlation between high TAMs infiltration and reduced OS in TP53^Mut patients with UC receiving platinum-based chemotherapy (figure 3E,F, TP53^Mut: median OS: 24 months vs not reached, log-rank p=0.050).

Figure 3. High tumor-associated macrophages infiltration is associated with chemotherapy- and immunotherapy-refractory outcome in p53-aberrant urothelial carcinoma. (A–B) Kaplan-Meier curves of OS stratified by TAMs infiltration in patients with UC on ACT in (A) the overall ZSHS cohort and (B) the p53^WT and p53^Abn subgroups. (C–D) Kaplan-Meier curves of OS stratified by TAMs infiltration in patients with UC on ACT in (C) the overall FUSCC cohort and (D) the p53^WT and p53^Abn subgroups. (E–F). Kaplan-Meier curves of OS stratified by TAMs infiltration in patients with UC on ACT in (F) the overall TCGA cohort and (F) the TP53^WT and TP53^Mut subgroups. (G–H). Kaplan-Meier curves of OS stratified by TAMs infiltration in patients with mUC on atezolizumab in (G) the overall IMvigor210 cohort and (H) the TP53^WT and TP53^Mut subgroups. Abn, abnormal; ACT, adjuvant chemotherapy; FUSCC, Fudan University Shanghai Cancer Center; mOS, median OS; mUC, metastatic urothelial carcinoma; Mut, mutant; OS, overall survival; TAMs, tumor-associated macrophages; TCGA, The Cancer Genome Atlas; UC, urothelial carcinoma; WT, wild type; ZSHS, Zhongshan Hospital Affiliated to Fudan University.

We then explored the association between TAMs infiltration and anti-programmed death-1 (PD-1)/programmed cell death-ligand 1 (PD-L1) immunotherapeutic response in UC, stratified by TP53 status. In the IMvigor210 cohort, high TAMs infiltration was linked to inferior OS in TP53^Mut patients treated with atezolizumab, while no such association was observed in TP53^WT patients (figure 3G,H, TP53^Mut: median OS: 7 vs 13 months, log-rank p=0.030). Importantly, multivariable Cox regression analysis revealed a statistically significant interaction between TP53 status and TAMs infiltration in predicting OS (online supplemental figure 2A, p for interaction=0.039), suggesting that the prognostic impact of TAMs may be modified by the aberrant p53 status. Specifically, multivariable Cox regression analysis indicated that high TAMs infiltration was independently associated with poor prognosis in the TP53^Mut subgroup (adjusted HR 1.908, 95% CI 1.177 to 3.093, p=0.009) after adjusting for established confounders including Eastern Cooperative Oncology Group (ECOG) performance status, PD-L1, and tumor mutational burden. In contrast, high TAMs infiltration did not show an independent prognostic association in the TP53^WT subgroup (online supplemental figure 2B).

Aberrant p53 status is linked to a lipid-metabolic shift and immunosuppressive polarization of TAMs

Given the clinical association between p53 status and protumorigenic TAM phenotypes, we interrogated functional programs of TAM subsets in the scRNA-seq dataset. Reg-TAMs exhibited enhanced immunosuppressive signatures compared with other subsets (figure 4A,B and online supplemental figure 3A,B). Gene Ontology analysis revealed that hypoxia, and metabolism-related pathways were upregulated in Reg-TAMs (figure 4B and online supplemental figure 3B). Trajectory analysis suggested that inflammatory cytokine-enriched TAMs (Inflam-TAMs) transitioned from lipid-associated TAMs (LA-TAMs) toward IFN-TAMs over pseudotime in p53^WT UC, whereas in p53^Abn UC, Inflam-TAMs appeared to shift from LA-TAMs toward Reg-TAMs (figure 4C,D and online supplemental figure 3C,D). Supporting this phenotypic transition, metabolic profiling via scMetabolism revealed that the trajectory-associated Reg-TAM and LA-TAM populations exhibited the highest overall metabolic pathway activities among all subsets. This hypermetabolic state was consistently observed in both the discovery and validation cohorts, indicating active metabolic reprogramming during the differentiation process (online supplemental figure E,F). In p53^WT UC, LA-TAMs were enriched for pro-inflammatory, antigen presentation, phagocytosis, and lipid storage pathways, while those in p53^Abn UC were enriched for immune suppression, fatty acid oxidation and lipid catabolism pathways (figure 4E,F and online supplemental figure 3G). In addition to the lipid-associated features observed in LA-TAMs, we further validated the molecular identity of other TAM subsets. While Reg-TAMs and IFN-TAMs were distinctly marked by high expression of SPP1 and CXCL9/CXCL10, respectively, we notably observed that Reg-TAMs also exhibited elevated levels of GPNMB, a canonical marker of LA-TAMs (online supplemental figure 3H). Bulk transcriptomic analysis further supported these divergent TAM phenotypes in TP53^WT versus TP53^Mut tumors in the TCGA cohort (figure 4G).

Figure 4. Aberrant p53 status is linked to a lipid-metabolic shift and immunosuppressive polarization of TAMs. (A) Box plots illustrating immunosuppressive gene signature scores across TAM clusters in UC samples (n = 23). The Reg-TAMs subpopulation serves as the reference group, with its median score indicated in gray. (B) Dot plot showing differentially enriched pathways (GO biological processes) between Reg-TAMs and other TAM clusters. (C–D) UMAP plots of Monocle3-derived TAM trajectories in (C) p53^WT and (D) p53^Abn tumors. (E–F) Bar plots demonstrating differentially enriched pathways (GO biological processes) in LA-TAMs compared with other TAM clusters within (E) p53^WT and (F) p53^Abn tumors. (G) Heatmap displaying signature scores for TAMs with different features between TP53^WT and TP53^Mut subgroups in the TCGA cohort. (H–I) Association between p53 status and TAM phenotypes in (H) the ZSHS IHC cohort and (I) the Shanghai FCM cohort. Abn, abnormal; Angio-TAMs, pro-angiogenic TAMs; FCM, flow cytometry; GO, Gene Ontology; IFN-TAM, interferon TAM; IHC, immunohistochemistry; IL, interleukin; Inflam-TAM, inflammatory cytokine-enriched TAMs; LA-TAM, lipid-associated TAMs; LAP, latency-associated peptide; Mut, mutant; PD-L1, programmed cell death-ligand 1; Reg-TAM, regulatory TAM; RTM-TAM, resident-tissue macrophage TAM; TAMs, tumor-associated macrophages; TCGA, The Cancer Genome Atlas; UC, urothelial carcinoma; UMAP, Uniform Manifold Approximation and Projection; WT, wild type; ZSHS, Zhongshan Hospital Affiliated to Fudan University. *p<0.05; ***p<0.001.

To provide orthogonal protein-level validation of these transcriptomic findings, we performed IHC and FCM analyses to evaluate the immunosuppressive features of TAMs from patients with UC stratified by p53 status. IHC analysis demonstrated a higher density of TAMs infiltrating the p53^Abn tumor microenvironment that expressed key functional molecules, including M2-like marker CD206 (p=0.004) and interleukin (IL)-10 (p=0.043), compared with p53^WT cases (figure 4H). Complementing these histological observations, FCM further quantified the expression levels of various immunoregulatory molecules within the TAM population. We observed a significant upregulation in the protein expression of multiple suppressive factors, including CD206 (p=0.028), IL-10 (p=0.005), latency-associated peptide (p=0.004), PD-L1 (p=0.026), and galectin-9 (p=0.008) in TAMs derived from p53^Abn tumors (figure 4I and online supplemental figure 4). The proportion of DC-SIGN-expressing TAMs also showed an upward trend in p53^Abn tumors, though without reaching statistical significance (p=0.178; online supplemental figure 4).

Aberrant p53 status characterizes a unique immune ecosystem of Reg-TAMs and suppressive CD4⁺ T cells in urothelial carcinoma

To explore TAM-mediated effects on the UC tumor immune microenvironment, we analyzed additional immune cell populations. In both the discovery and validation cohorts, p53^Abn UC exhibited increased CD4⁺ T cells infiltration compared with p53^WT UC, whereas no marked differences were observed in the infiltration levels of CD8⁺ T cells or other immune lineages (figure 5A and online supplemental figure 5A). Subclustering of CD4⁺ T cells consistently identified major functional subsets, including CD4T_Activated, CD4T_Cyto _GZMA, CD4T_Treg, and CD4T_Tex_LAG3, across both cohorts (figure 5B, online supplemental figure 5B,D,E). While p53^WT UC was characterized by activated and cytotoxic CD4⁺ T cells (CD4T_Activated, CD4T_Cyto), p53^Abn UC was enriched for suppressive subsets (CD4T_Treg, CD4T_Tex_LAG3, and CD4T_Dysfunction_MT1E) (figure 5C and online supplemental figure 5C). Bulk GSEA demonstrated upregulation of immune suppression and regulatory T cell (Treg) differentiation pathways in TP53^Mut UC compared with TP53^WT UC in the TCGA cohort (figure 5D). IHC analysis corroborated these findings, demonstrating higher densities of total CD4⁺ T cells and Tregs in p53^Abn UC (figure 5E). FCM analysis further revealed that the abundance of both total CD4⁺ T cells and Tregs was positively correlated with TAMs infiltration specifically in p53^Abn UC (figure 5F). Notably, TAMs infiltration correlated with cytotoxic CD4⁺ T cells in p53^WT UC, whereas in the p53^Abn context, it correlated with proliferative and exhausted CD4⁺ T cells (figure 5G). In the ZSHS cohort, p53^Abn tumors with high CD4⁺ T cells infiltration exhibited increased densities of cells expressing exhaustion markers, including cytotoxic T-Lymphocyte-associated protein 4 (CTLA-4), T-cell immunoglobulin-domain and mucin-domain-containing molecule 3, PD-1, and T-cell immunoreceptor with immunoglobulin and immunoreceptor tyrosine-based inhibitory motif (figure 5H), with similar trends in the overall p53^Abn subgroup (online supplemental figure 5F).

Figure 5. Aberrant p53 status characterizes a unique immune ecosystem of Reg-TAMs and suppressive CD4⁺ T cells in urothelial carcinoma. (A) Pie charts depicting the proportion of major immune cell lineages in p53^WT (top) and p53^Abn (bottom) tumors. (B) UMAP plots visualizing the distinct subclusters of CD4⁺ T cells. (C) Pie charts depicting the proportion of CD4⁺ T-cell subsets in p53^WT (top) and p53^Abn (bottom) tumors. (D) GSEA demonstrating the enrichment of immunosuppressive and Treg-related GO pathways in TP53^Mut compared with TP53^WT tumors in the TCGA cohort. (E–F) Association between p53 status and the infiltration levels of CD4⁺ T cells and Tregs in (E) the ZSHS IHC cohort and (F) the Shanghai FCM cohort. (G) Heatmap of Spearman correlation coefficients between CD68⁺ TAM abundance and the functional states (proliferation, cytotoxicity, and exhaustion) of CD4⁺ T cells, stratified by p53 status in the Shanghai FCM cohort. (H) Quantification of immune checkpoint-positive cells in the ZSHS IHC cohort, stratified by both p53 status and CD4⁺ T-cell infiltration levels. (I) Dot plot visualizing the significant ligand-receptor interactions inferred by CellChat between Reg-TAMs and suppressive CD4⁺ T-cell subsets specifically in p53^Abn tumors. (J) Comparison of the Reg-TAM/suppressive CD4⁺ T-cell interaction signature score between TP53^Mut and TP53^WT tumors in the TCGA (top) and IMvigor210 (bottom) cohorts. Abn, abnormal; CTLA-4, cytotoxic T-Lymphocyte-associated protein 4; FCM, flow cytometry; GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; IHC, immunohistochemistry; Mut, mutant; NES, normalized enrichment score; NK, natural killer; PD-1, programmed death-1; Reg-TAM, regulatory TAM; TAMs, tumor-associated macrophages; TCGA, The Cancer Genome Atlas; TIGIT, T-cell immunoreceptor with immunoglobulin and immunoreceptor tyrosine-based inhibitory motif; TIM-3, T-cell immunoglobulin-domain and mucin-domain-containing molecule 3; Treg, regulatory T cell; UMAP, Uniform Manifold Approximation and Projection; WT, wild type; ZSHS, Zhongshan Hospital Affiliated to Fudan University.

The observed co-infiltration of Reg-TAMs and suppressive CD4⁺ T cells in p53^Abn UC suggested possible intercellular crosstalk. We then applied CellChat to infer receptor-ligand interactions between Reg-TAMs and suppressive CD4⁺ T cells. Reg-TAMs were predicted to signal predominantly via the SPP1-CD44 axis and interact through CD86-CTLA-4 signaling in both the discovery and validation cohorts (figure 5I and online supplemental figure 5G). Although these in silico predictions do not confirm direct interactions, they delineate a potential regulatory model in which Reg-TAMs contribute to suppressive CD4⁺ T-cell phenotypes in p53^Abn UC. Consistently, the Reg-TAM/suppressive CD4⁺ T-cell interaction signature was upregulated in TP53^Mut UC across both the TCGA and IMvigor210 cohorts (figure 5J). Furthermore, transcriptomic correlation analyses in the TCGA cohort revealed a significant positive association between the SPP1⁺ TAM signature and CD44 expression (online supplemental figure 6A). Crucially, the elevated SPP1-CD44 axis signature was strongly correlated with higher exhaustion scores in CD4⁺ T cells (online supplemental figure 6B,C). Together, these multimodal computational and orthogonal validations support the hypothesis that the SPP1-CD44 axis may contribute to the suppressive immune microenvironment in p53-aberrant UC.

Discussion

Given the pivotal role of TP53 in tumor immunity²⁰ and the pronounced enrichment of immunosuppressive TAMs in UC, we aimed to investigate the dynamic interplay between p53 status and TAMs in UC. By leveraging single-cell transcriptomics, we uncovered distinct TAM landscapes associated with p53 status that correlate with patient prognosis and therapy response across multiple cohorts. The prognostic impact of TAMs was statistically robust in the FUSCC and TCGA cohorts, a trend consistently observed in the ZSHS dataset. These collective findings across independent cohorts underscore the reliable association between high TAMs infiltration and adverse clinical outcomes within the p53-aberrant microenvironment.

Our integrative multi-omics analysis aligns with emerging frameworks describing TAMs diversity across tumors.²¹ Most canonical TAM subsets were recapitulated in UC, with notable polarization differences by p53 status. IFN-TAMs, marked by CXCL9, CXCL10, CXCL11 and IDO1, were enriched in p53^WT UC, suggesting intact interferon signaling. In contrast, p53^Abn UC were dominated by Reg-TAMs expressing immunosuppressive markers such as SPP1, MARCO, SLC2A1, and INHBA. The opposing expression of CXCL9 and SPP1 highlights a binary polarization axis within the TIME.²²

Beyond immunoregulation, Reg-TAMs in p53^Abn UC exhibited enhanced adaptation to hypoxia through upregulated lipid metabolic programs. We also identified an LA-TAMs subset, expressing TREM2, GPNMB, APOC1 and APOE, present across both p53^Abn and p53^WT subtypes, though not differing significantly in abundance. Lipid metabolism in macrophages has context-dependent roles: lipid catabolism tends to support immunosuppression, while lipid synthesis can promote inflammation.^{23 24} We hypothesize that LA-TAMs represent a transitional state, differentiating toward either immunosuppressive or pro-inflammatory phenotypes depending on the p53 context. Notably, LA-TAMs in p53^WT tumors were characterized by phagocytosis, lipid storage, antigen presentation, and pro-inflammatory signaling, whereas those in p53^Abn tumors were skewed toward fatty acid oxidation. This suggests that while lipid accumulation is a shared feature, the downstream metabolic fate, specifically storage versus catabolism, may diverge based on the p53 status, coinciding with distinct immunoregulatory outcomes. Trajectory analyses revealed that in p53^WT UC, Inflam-TAMs progressed toward IFN-TAMs via LA-TAMs, whereas in p53^Abn tumors, Inflam-TAMs were reprogrammed into Reg-TAMs through LA-TAMs. This suggests that aberrant p53 status is associated with metabolic reprogramming and the polarization of TAMs toward immunoregulatory fates.

Crosstalk between TAMs and tumor cells has been observed in other cancers, such as hepatocellular carcinoma²⁵ and glioma,²⁶ where tumor-driven lipid remodeling maintains TAMs survival and facilitates immune suppression. LA-TAMs, like TREM2⁺ macrophages, long associated with metabolic homeostasis,²⁷ exhibit context-dependent roles in cancer. While they perform protective functions in early disease (eg, MASLD/MASH), PPARγ-driven terminal differentiation can shift their phenotype toward protumorigenic states via MMP9 and SPP1 upregulation.²⁸ IL-10 is another key immunosuppressive regulator modulated by metabolic rewiring.^{29 30} We observed high infiltration of IL-10⁺ TAMs in p53^Abn UC, consistent with prior findings that p53 inactivation drives M2-like polarization, ultimately enabling immune evasion through CD8⁺ T-cell suppression.³¹ In addition, p53-aberrant tumors have been reported to exhibit tissue remodeling and ecosystem-level changes associated with enrichment of protumor macrophage states (eg, SPP1⁺ TAMs), which may further reinforce macrophage metabolic adaptations.³²

Macrophages orchestrate both upstream immune sensing and downstream effector responses. Established evidence showed that while CD8⁺ T cells show limited tumor-specific remodeling in UC, CD4⁺ T cells display considerable diversity.³³ Tumor-infiltrating cytotoxic CD4⁺ T cells, when not inhibited by Tregs, can mediate antitumor responses and predict immunotherapy efficacy.^{33 34} In our study, p53^Abn tumors harbored fewer cytotoxic CD4⁺ T cells and more suppressive subsets (eg, LAG-3⁺, CTLA-4⁺, and metallothionein-expressing CD4⁺ T cells), alongside increased Tregs infiltration. TAMs were also found to co-infiltrate with exhausted CD4⁺ T cells and Tregs in p53^Abn tumors. Mechanistically, our in silico interaction analysis predicts the SPP1-CD44 axis as a prominent communication pathway between Reg-TAMs and suppressive CD4⁺ T cells in p53^Abn tumors. We hypothesize that this axis, potentially synergizing with the CD86-CTLA4 checkpoint signal, reinforces T-cell exhaustion. While these transcriptomic predictions strongly suggest a coordinated immunosuppressive network, we acknowledge the current lack of direct spatial co-localization evidence. Therefore, this SPP1-CD44 interaction represents a hypothesis-driven model that warrants future experimental and spatial validation. Nevertheless, this putative model aligns with previous studies showing that SPP1 secreted from TAMs promotes tumor cell invasiveness and suppresses CD8⁺ T cells responses, especially in metastases.^{35 36} Thus, aberrant p53 status appears to be linked to a global immune skewing in the UC TIME through TAMs polarization and myeloid-T cell crosstalk.

Several limitations should be acknowledged. First, although our multicohort human scRNA-seq analyses consistently associate tumor p53 status with distinct TAM polarization and lipid metabolic programs, definitive causality remains to be established through direct functional perturbation models. Second, we acknowledge a methodological distinction between p53 IHC for in-house stratification and TP53 sequencing for public datasets. Although this introduces platform-specific variability, these two modalities exhibit high clinical concordance and yield consistent immune associations across all independent cohorts. Such cross-platform reproducibility reinforces the biological robustness of our findings within the p53-aberrant context. Third, the proposed SPP1-CD44 interaction between Reg-TAMs and T cells is based on computational inference and therefore requires direct spatial and functional validation. Together, these findings provide a compelling foundation for future mechanistic studies in p53-aberrant UC.

In summary, this study revealed that aberrant p53 status in UC is closely linked to immune ecosystem remodeling, particularly through TAMs reprogramming. Aberrant p53 status correlated with the emergence of metabolically active, immunosuppressive macrophage subsets that interact with suppressive CD4⁺ T cells and Tregs via the SPP1-CD44 axis, shaping a TIME resistant to immune clearance and therapy. These findings provide mechanistic insight into p53-associated immune evasion and suggest new strategies for therapeutic stratification based on tumor genomic and immune contexture.

Data availability statement

Data are available upon reasonable request.