Alexandre Vallée
Alexandre Drezet3
Maxence Arutkin- 1Department of Epidemiology and Public Health, Foch Hospital, Suresnes, France
- 2Département Universitaire de Santé Publique, Prévention, Observation, Territoires (SPOT), University of Versailles, Saint-Quentin-en-Yvelines (UVSQ), Versailles, France
- 3Direction de la Recherche Clinique et de l’Innovation (DRCI), Foch Hospital, Suresnes, France
- 4School of Chemistry, Center for the Physics and Chemistry of Living Systems, Tel Aviv University, Tel Aviv-Yafo, Israel
Mass cytometry (CyTOF) and Imaging Mass Cytometry (IMC) provide single-cell resolution for over 50 protein markers, enabling unprecedented exploration of tumour and immune heterogeneity. We conducted a scoping review of 61 original studies (inception–2025), spanning 17 cancer types, to map current applications, analytical strategies, and emerging biological insights. 46 studies used CyTOF alone, 12 employed IMC exclusively, and 3 combined both platforms. Median panel sizes were 33.5 markers for CyTOF and 33 for IMC. While lineage and immune checkpoint markers were universal, phospho-epitopes, metabolic enzymes, and stromal proteins appeared in more focused subsets. Most studies followed a three-step analytical workflow: (i) segmentation or gating, (ii) unsupervised clustering, and (iii) downstream spatial or functional analyses. CyTOF investigations frequently identified exhausted CD8+ T-cell subsets (e.g., PD-1+TIM-3+CD39+), suppressive myeloid populations (e.g., CD163+HLA-DR− macrophages), and metabolically reprogrammed Tregs. IMC studies uncovered spatial patterns predictive of outcome, such as tertiary lymphoid structures (TLSs) and macrophage–T cell exclusion zones. Several studies proposed predictive immune signatures or integrated CyTOF with transcriptomic or spatial datasets. We identified five recurrent immunobiological motifs, CD8+ T-cell bifurcation, CD38+ TAM barriers, TLS maturity, CTLA-4+ NK-cell signatures and metabolically defined niches, highlighting convergent axes of resistance and response. Bioinformatic pipelines converged around FlowSOM or PhenoGraph clustering, CITRUS or elastic-net feature selection, and increasingly, machine learning and agent-based spatial modelling. Collectively, CyTOF and IMC are redefining biomarker discovery, therapeutic stratification, and virtual trial design in oncology, establishing high-dimensional CyTOF as a cornerstone of next-generation precision cancer medicine.
1 Introduction
Cancer is marked by remarkable cellular heterogeneity: malignant clones coexist with diverse immune, stromal and vascular elements that continually evolve under therapeutic pressure (1–3). Conventional bulk “omics” approaches obscure this complexity, whereas single-cell technologies now allow high-resolution dissection of the tumour microenvironment (TME) (4). Among these, mass cytometry, commercially known as Cytometry by Time-Of-Flight (CyTOF), has emerged as a leading platform for multidimensional profiling of individual cells in both suspension and tissue sections (5–7).
CyTOF couples inductively coupled-plasma ionisation with time-of-flight mass spectrometry to quantify antibodies conjugated to isotopically pure heavy-metal tags. By eliminating the spectral overlap inherent to fluorescence, a single assay can routinely measure > 50 protein markers per cell without the need for compensation, while maintaining flow-cytometry-level throughput and sensitivity (8, 9). This capacity enables simultaneous assessment of lineage, activation, signalling and metabolic markers, delivering a systems-level view of immune and tumour cell states.
Since its introduction, CyTOF has transformed immuno-oncology research. High-parameter phenotyping has revealed treatment-induced remodelling of peripheral and intratumoural immunity, uncovered rare suppressive or progenitor T-cell subsets linked to checkpoint-blockade outcomes, and identified circulating signatures that anticipate toxicities or relapse (5, 10). Furthermore, integration with transcriptomics and machine-learning pipelines is accelerating the translation of CyTOF-derived biomarkers toward personalised therapy and minimal-residual-disease monitoring in clinical trials (11–13).
An important derivative technology, Imaging Mass Cytometry (IMC), extends the CyTOF principle to formalin-fixed tissue sections via laser ablation and metal-tag detection, preserving spatial context at sub-cellular resolution (14, 15). IMC has already mapped immunological “hot” and “cold” niches in melanoma, quantified tertiary lymphoid structures in hepatobiliary tumours and generated spatial risk scores that outperform conventional histopathology (16).
Given the rapid expansion of CyTOF and IMC applications (17), a comprehensive synthesis is timely. This scoping review collates original studies spanning various malignancies, summarises panel design trends, computational pipelines and key biological insights, and highlights emerging opportunities for biomarker discovery and therapeutic stratification in oncology.
2 Methods search strategy
A comprehensive literature search was conducted across three databases, PubMed/MEDLINE, EMBASE, and Science Direct, to capture relevant studies published from inception up to April 30, 2025.
The search strategy combined keywords related to CyTOF with terms related to cancer. In each database, we used Boolean operators to broaden or restrict the search as appropriate. For example, in PubMed the following keywords were used: (“CyTOF” OR “mass cytometry”) AND (“cancer” OR “tumour” OR “neoplasm”) AND (“biomarker” OR “signature” OR “immune profile”) AND (“prediction” OR “predictive” OR “prognostic” OR “diagnostic” OR “response”).
This strategy was adjusted to each database’s syntax. In EMBASE and Science Direct, we searched within article Title/Abstract/Keywords for terms. No language filter was applied initially, but we later limited inclusion to English-language results.
2.1 Eligibility criteria
Studies were eligible for inclusion if they:
1. were original research articles (including experimental studies, observational studies, computational modelling studies) published in peer-reviewed sources;
2. focused on a CyTOF/IMC application in the context of cancer;
3. were published in English; and
4. were published between inception and April 2025 (inclusive).
We excluded review articles, meta-analyses, pre-prints, editorials/commentaries, and conference abstracts without full papers from the main analysis, since our focus was on original data. Where eligibility was unclear, we reached consensus among reviewers by carefully assessing whether the study’s methods aligned with the topic.
2.2 Study selection
All references retrieved from the database searches were imported into a reference manager, and duplicates were removed. Title and abstract screening was performed on the unique records to identify obviously irrelevant papers. Two reviewers independently screened each title/abstract against the eligibility criteria. We retained any article that either reviewer marked as potentially relevant. Next, we obtained the full-text of all remaining articles and conducted a detailed evaluation for final inclusion. Disagreements on inclusion were resolved through discussion or consultation with a third reviewer. The selection process was documented following PRISMA guidelines.
The initial search across databases yielded a total of 148 records. Of them, 51 were excluded based on title/abstract screening. We assessed full-text articles for eligibility. 36 were excluded for reasons such as not being original research (e.g., review), not actually involving the topic of the research, or being commentary/case report. 61 studies met all criteria and were included in the qualitative synthesis and data extraction for this review (Figure 1). Supplementary Table S1 presents the characteristics of all the 61 analysed studies.
Figure 1. Flow diagram.
2.3 Risk of bias assessment
To assess the methodological quality and potential risk of bias of the included studies, we applied a structured framework adapted from established tools depending on study type. For observational studies (retrospective or prospective cohorts), we used a simplified version of the NIH Quality Assessment Tool (18), focusing on five core criteria: clearly defined population, justification of sample size, blinding of outcome assessors, validity of outcome measurements, and presence of external validation. For interventional studies without randomisation, we referenced ROBINS-I (Risk Of Bias in Non-Randomized Studies – of Interventions) (19) to evaluate confounding and selection bias. For studies involving diagnostic marker evaluation (e.g., spatial IMC biomarkers), we used selected domains from QUADAS-2 (20). Preclinical studies were noted as not applicable for formal bias scoring but were qualitatively appraised. Each study received a global bias rating (low, moderate, or high) based on the number of criteria met and methodological transparency. The full bias matrix with per-criterion assessments and assigned method is provided in Supplementary Table S2.
3 Results
3.1 Study characteristics
The 61 eligible papers span 17 malignant entity types and remain dominated by suspension CyTOF (46/61, 75%) (21–66), with IMC-only analyses (12/61, 20%) (67–78) and dual-platform studies (3/61, 5%) making up the remainder. Solid tumours account for 52 of 61 studies (85%) (79–81), while haematological malignancies (acute leukaemias and lymphomas) contribute 9 (15%). The most represented families were Skin/Melanoma (n=10), Lung/Thoracic (n=9), Leukaemia/Lymphoma (n=9), Liver/Hepatic (n=7), Bladder/Urothelial (n=7), and Colorectal/Colon/Rectal (n=6). Mid-frequency families included Breast (n=4) and Pancreatic (n=3), while Esophageal/Gastric (n=3) and Gynaecological (n=3) were also present. The remaining families, Head & Neck (n=2), General carcinoma/other (n=2), Mesothelioma (n=2), Renal/Kidney (n=2), Sarcoma (n=2), Brain/CNS (n=1), and Prostate (n=1), were less frequent. Counts are per family; multi-cancer studies contribute to multiple families (Supplementary Table S1).
3.2 CyTOF and IMC panel design and marker utilisation
Across the 61 eligible entries, we catalogued 46 CyTOF articles and 12 IMC articles, with 3 studies deploying both technologies. Although panel composition was highly study-specific, three clear trends emerged (Table 1).
Table 1. Summary of CyTOF and IMC panel characteristics.
3.2.1 Lineage-based CyTOF cores supplemented by context-specific functional markers
The median CyTOF panel contained 35 antibodies (std 9.6). CyTOF panels typically include broad immune lineages: 41% contain the full CD3/CD4/CD8/CD19/CD56/CD14 backbone, and 73% include ≥ 1 immune-checkpoint (PD-1/TIM-3/TIGIT/CTLA-4).
Functional markers were then tailored to the biological question:
- DNA-damage sensors, for example p53 and γH2AX, were added to a glioblastoma drug-response assay to track chemo-induced stress (26).
- For example, Transferrin receptor (TFRC) was incorporated to monitor iron metabolism in an anlotinib + PD-1 hepatocellular-carcinoma study (23).
- For example, CD39 and CD103 were paired to discriminate Texstem from Texterm CD8 subsets in malignant pleural effusions (54).
3.2.2 IMC panels combining tumour, immune and stromal features
IMC panels median 33.0 antibodies (std 8.0) and usually blended immune-lineage markers with stromal or extracellular-matrix components (e.g., α-SMA, collagen-I, FAP), enabling simultaneous cellular and spatial read-outs.
- For example, an oesophageal-squamous-cell-carcinoma study deployed a 25-marker stromal-centric panel to map CAF–TAM co-localisation (68).
- For example, the NeoTRIP trial in triple-negative breast cancer used a 43-marker pan-immune/tumour panel that included CD15 to highlight a chemoresistant tumour subset (67).
3.3 Marker reuse and cross-study convergence
25 antibodies appeared in more than 70% of 17 types of cancers, more than 35% of panels and both in the two technics (CyTOF and IMC) among the 61 studies (Table 2). Figure 2 shows the frequency distribution of the 25 core immune markers grouped by biological class (lineage, checkpoint, Treg, myeloid, proliferation, etc.). This treemap highlights the dominance of T-cell and checkpoint markers across panels, while also illustrating the recurrent inclusion of myeloid, B-cell, NK and APC/MHC-II markers.
Table 2. Core immune markers across 61 studies for panel convergence and their associated metabolic axes and functional pathways.
Figure 2. Core immune marker frequencies grouped by biological class across CyTOF/IMC studies. Treemap representation of the 25 most frequently used immune markers identified across 61 CyTOF and IMC studies, grouped by functional and biological class. Each box corresponds to a marker, with its relative size reflecting frequency of use across cancer types and study panels. Major classes include checkpoint receptors (PDCD1/PD-1, CD274/PD-L1, CTLA-4, HAVCR2/TIM-3), T-cell lineage markers (CD3E, CD4, CD8A), regulatory T-cell (Treg) markers (FOXP3, IL2RA), memory/homing markers (IL7R, CCR7), B-cell and NK-lineage markers (CD19, NCAM1/CD56), and myeloid subsets (CD14, ITGAM, ITGAX, CD68). Additional categories capture activation (CD27, ICOS), proliferation (MKI67), cytotoxicity (GZMB), adhesion/stemness (CD44), and APC/MHC-II molecules (HLA-DRA/DRB1). This functional grouping highlights the convergent design of CyTOF and IMC panels across tumour types, where T-cell and checkpoint markers dominate, but recurrent inclusion of myeloid, NK, B-cell and antigen-presentation modules illustrates the multi-compartmental view required to capture tumour–immune interactions.
Across the 61 studies analysed, we observed a convergent panel of immune markers that consistently map onto five major metabolic axes of the tumour microenvironment:
- Glycolysis (Warburg metabolism),
- Oxidative phosphorylation (OXPHOS),
- Fatty acid oxidation (FAO)/lipid metabolism,
- Amino acid metabolism (arginine, glutamine, tryptophan, NAD+/adenosine),
- Nucleotide biosynthesis.
These axes capture the central modes of energy and biosynthetic adaptation that govern immune cell function, plasticity, and exhaustion in cancer. Glycolysis represents the classical “Warburg” program, fuelling rapid glucose uptake and lactate production to sustain proliferating and cytotoxic effector T cells as well as NK cell activity (82, 83). Oxidative phosphorylation (OXPHOS) reflects mitochondrial respiration via the tricarboxylic acid cycle, which is critical for memory T cells and long-lived plasma cells, ensuring durable immunity (84). Fatty acid oxidation (FAO) and broader lipid metabolism dominate in regulatory T cells and M2-like macrophages, linking β-oxidation to immune suppression and tissue repair (85). Amino acid metabolism, including arginine, glutamine, and tryptophan pathways, regulates effector responses and checkpoint-associated exhaustion, with enzymes such as Arg1 and IDO shaping the immune landscape through nutrient depletion and metabolite production (86). Finally, nucleotide biosynthesis underpins cell cycle progression, antibody production, and clonal expansion, as proliferating Ki-67+ lymphocytes rely on enhanced purine and pyrimidine synthesis (87, 88).
Figure 3 further illustrates the prevalence of these 25 core immune markers across cancer families. While T-cell lineage and checkpoint markers were universally represented, myeloid and stromal components showed more heterogeneous inclusion, highlighting disease-specific tailoring of CyTOF/IMC panels.
Figure 3. Per-family prevalence of core immune markers across CyTOF/IMC studies. Heatmap showing the prevalence of the 25 core immune markers (x-axis) across cancer families (y-axis), calculated as the fraction of study panels in which each marker was included. Warmer colors (yellow) denote high prevalence (present in nearly all panels for a given cancer family), whereas cooler colors (green to purple) indicate more restricted or cancer-specific inclusion. T-cell lineage markers (CD3E, CD4, CD8A) and checkpoint molecules (PDCD1/PD-1, CD274/PD-L1, CTLA-4, HAVCR2/TIM-3) were consistently present across nearly all families. In contrast, myeloid-associated markers (CD14, ITGAM, ITGAX, CD68), NK-cell (NCAM1/CD56), and TLS-related or activation markers (ICOS, CCR7, IL7R) showed more variable inclusion, reflecting cancer-type specific emphasis. This visualization underscores the convergent yet context-dependent design of CyTOF and IMC panels, where universal backbones (lineage and checkpoints) are supplemented by lineage- or disease-tailored modules to capture relevant tumour–immune biology.
3.4 Tumour and immune cell populations identified
CyTOF and IMC profiling resolved five recurrent cellular modules:
i. CD8/CD4 T-cell differentiation and exhaustion states,
ii. B-cell and tertiary-lymphoid-structure programmes,
iii. innate-lymphoid and NK-cell subsets,
iv. myeloid-lineage heterogeneity, and
v. tumour–stromal phenotypes (Table 3).
Table 3. Key cell populations discovered across the studies.
The most frequently highlighted phenotype was the exhausted PD-1hi TIM-3+ CD39+ CD8+ T-cell subset, detected in 9 of the 61 studies (14%) and typically enriched among non-responders, e.g. in peripheral blood of advanced NSCLC patients (47) and in cases of immune-checkpoint-inhibitor pneumonitis (59). Conversely, tissue Abbreviations: APC, A-resident memory CD8 T cells (CD103+ CD69+) and stem-like Texstem cells (PD-1int CD39− CD28+) tracked durable benefit, as shown in endometrial cancer treated with cabozantinib + nivolumab (39) and in malignant pleural effusions from thoracic tumours (54).
Myeloid diversity was equally striking. Urothelial-carcinoma specimens contained CD38+ tumour-associated macrophages (TAMs) with stronger immunosuppressive signatures than canonical PD-L1+ TAMs (49), whereas diffuse large B-cell lymphoma blood samples showed expansion of prognostically adverse S100A9high non-classical monocytes (56).
Mapping of the innate-lymphoid compartment uncovered disease-specific NK-cell states: CD56bright CD69+ activated NK cells were enriched in high-risk acute lymphoblastic leukaemia (48), while a conserved CTLA-4+ NK signature predicted ipilimumab responsiveness across mouse and human melanomas (52).
Finally, several studies focused on tumour-intrinsic and stromal programmes. In lung adenocarcinoma, a NOTCH3-driven network of FAP+ cancer-associated fibroblasts and ACTA2+ pericytes delineated perivascular immune-exclusion zones (35). Bladder tumours contained IGF2BP3high ALDH+ cancer stem-like cells that shaped responsiveness to anti-PD-1 therapy (80), while prostate tumours harboured rare CD15+ epithelial subclones linked to high-grade pathology (29).
3.5 Bioinformatics tools and algorithms utilised
Analytical workflows varied substantially across the 61 studies, yet several core toolsets recurred consistently (Table 4). Dimensionality-reduction and clustering constituted the entry point for virtually every CyTOF or IMC dataset. Across the corpus, viSNE/t-SNE projections and UMAP embeddings were the most frequently adopted two-dimensional visualisations, while FlowSOM self-organising maps and PhenoGraph community detection dominated unsupervised clustering. These methods enabled, for example, construction of an immune atlas of urothelial carcinoma and classification of B-cell subtypes in lung adenocarcinoma.
Table 4. Key computational tools, frequency of use, and exemplar studies.
Supervised differential-abundance frameworks were less common but crucial when outcome-associated sub-populations were sought. CITRUS facilitated feature selection in metastatic pancreatic cancer, and CITRUS combined with elastic-net regularisation yielded blood-based predictive models in melanoma. Limma/edgeR-based diffcyt pipelines offered a parametric alternative.
Predictive modelling methods, a subset of machine-learning approaches, and cross-cohort transfer-learning are rapidly gaining traction. The random-forest-based hDirect-MAP algorithm successfully transferred T-cell phenotypes across five independent skin-cancer cohorts. CoGAPS factorisation coupled with projectR enabled cross-species learning between murine and human checkpoint-inhibitor studies.
Spatial IMC datasets exploited specialised graph analytics. Graph-based community detection informed a cabozantinib + nivolumab hepatocellular-carcinoma study, whereas nearest-neighbour distance scoring delineated stromal niches in oesophageal squamous-cell carcinoma. One study even integrated IMC with Visium spatial transcriptomics in an agent-based spQSP simulation to perform in silico virtual trials.
Panel-specific utilities also appeared. CyTOFmerge harmonised datasets with non-overlapping marker panels, and an automated batch-correction engine (e.g., Harmony or Seurat Integration) was reported in 29 studies.
A call for a unified analytical framework: The wide dispersion of tools, several studies relying on t-SNE while 31 omit any advanced batch integration, underscores a fragmentation that complicates cross-study synthesis. It argues for a unifying pipeline grounded in geometric scattering, entropic optimal-transport distances, and semi-supervised graph learning. Because the scattering representation is provably stable to affine batch perturbations and OT provides a principled geometry for whole-population comparison, this framework could replace the current patchwork of visualisation-first heuristics with an end-to-end, mathematically rigorous alternative. Adopting such a cohesive approach would directly address the two most frequent failure modes revealed by the audit, uncorrected batch effects and distortion-prone embeddings, while offering a common feature space that enables meta-analysis across future studies.
3.6 Spatial analysis by imaging mass cytometry
15 of the 61 unique studies (25%) generated IMC data, and many carried out explicit spatial statistics beyond single-cell quantification (Table 5).
Table 5. Spatial analysis strategies, frequency and exemplar findings.
Most groups applied a similar three-step workflow:
1. Sub-cellular segmentation of 18- to 46-plex images (median 33);
2. Phenotype assignment from multiplex marker intensities;
3. Distance- or graph-based calculations to capture cell–cell interactions and tissue organisation.
15 of the 61 studies generated IMC data, and many went beyond simple single-cell counts to apply explicit spatial statistics. Most groups followed the same three-step workflow: first, they segmented 25- to 43-plex images into single-cell masks; next, they assigned phenotypes from multiplexed marker intensities; and finally, they calculated distance- or graph-based metrics to quantify cell–cell interactions and tissue organisation.
Nearest-neighbour and perimeter metrics were reported in seven studies. In the NeoTRIP trial of triple-negative breast cancer, the average distance between proliferating CD8 TCF1-positive T cells and MHC-II–positive tumour cells predicted pathological complete response (67). In oesophageal squamous-cell carcinoma, a perimeter score that combined α-SMA–positive cancer-associated fibroblasts with CD163-positive macrophages stratified overall survival (68).
The majority of the studies modelled tissues as spatial graphs. For example, in hepatocellular-carcinoma biopsies from a cabozantinib plus nivolumab trial, modularity-based community detection separated “immune-inflamed” from “immune-excluded” regions; enrichment of a CD8-negative, Arg1-high macrophage module was linked to treatment resistance (69).
Two investigations performed whole-section archetype or niche mapping. A melanoma study defined six tumour-microenvironment archetypes ranging from myeloid-dense “immune-cold” to lymphoid-rich “immune-hot” niches, with the latter correlating with durable PD-1 benefit (76). In pancreatic ductal adenocarcinoma, co-registration of adenosine mass-spectrometry imaging with IMC revealed hypoxic extracellular-adenosine niches co-localised with Arg1-high macrophages and regulatory T cells (78).
Finally, several studies focused on tertiary lymphoid structures or immune micro-aggregates. In combined hepatocellular–cholangiocarcinoma, an intra-tumour TLS score derived from a 32-marker panel was associated with lower regulatory-T-cell density and improved overall survival (71). In HPV-positive oropharyngeal squamous-cell carcinoma, CXCL13-driven aggregates of CD8, CD4 and dendritic cells defined an “immune-reactive” subtype with favourable ten-year survival (77).
Beyond individual studies, we synthesised the spatial findings into three conserved archetypes, myeloid/TAM barriers, TLS maturity, and CD8 distribution, captured across multiple cancers and platforms.
3.7 Clinical impact and translational integration of CyTOF/IMC signatures
While CyTOF and IMC offer high-dimensional resolution of the tumour–immune landscape, their ultimate value lies in shaping clinical decision-making. Several studies have begun to demonstrate how specific cellular or spatial signatures derived from these platforms inform patient stratification, treatment selection, and response monitoring (Table 6).
Table 6. Translational highlights of CyTOF/IMC-derived biomarkers in oncology.
Expanding on these translational highlights, a recurrent theme is the bifurcation between progenitor and terminally exhausted CD8+ T cells. Progenitor Texstem subsets (TCF1+, PD-1 int, CD39−) consistently associate with durable ICI benefit across melanoma, NSCLC, and breast cancer, while Texterm (PD-1hi TIM-3+ CD39+) predict therapeutic failure. Similarly, myeloid suppressive circuits emerge as critical determinants of immune escape. CD38+ and Arg1+ TAMs create metabolic and spatial barriers, observed in HCC, PDAC, and urothelial carcinoma, often correlating with non-response to ICI. These findings support ongoing trials of anti-CD38 antibodies and arginase inhibitors in solid tumours. Spatial biomarkers derived from IMC further underscore the importance of tissue architecture. TLS maturity and CAF–TAM exclusion perimeters are not only descriptive, but provide quantifiable, reproducible metrics that stratify patient prognosis and response. Collectively, these insights indicate that CyTOF/IMC biomarkers do not merely catalogue immune states but increasingly guide patient selection, risk stratification, and rational design of combination therapies.
3.7.1 Stratification in immunotherapy trials
In melanoma and NSCLC, CyTOF-derived immune signatures have been used to predict differential response to anti-PD-1 or anti-CTLA-4 therapies. For instance, the frequency of stem-like PD-1 int CD39− CD28+ CD8+ T cells has emerged as a predictor of durable response to checkpoint blockade in both tumour tissue and effusion samples (53, 54, 67). Similarly, exhausted PD-1hi TIM-3+ CD8+ T cells are enriched in non-responders and have been used to define immunotherapy-resistant profiles (53, 57).
3.7.2 Spatial risk scores improving histopathology
In oesophageal squamous cell carcinoma, an “CAF–macrophage perimeter score” stratified overall survival independently of tumour stage (68), while in Hodgkin lymphoma, a composite spatial score based on CXCL13+ macrophage–CXCR5+ tumour cell pairing predicted post-transplant relapse with higher precision than any single marker (74).
3.7.3 Guiding combinatorial treatments
In hepatocellular carcinoma, CyTOF–IMC integration was used to identify Arg1hi macrophage-dense niches resistant to cabozantinib plus nivolumab, prompting spatial simulation (spQSP) of alternative regimens (70).
3.8 Emergent immunobiological motifs and hypothesis-generating convergence
Beyond cataloguing marker panels and cellular phenotypes, our synthesis identified several cross-study biological motifs that recur across distinct tumour types, suggesting deeper, conserved mechanisms of immune–tumour interaction. These motifs, summarised in Table 7, represent not just descriptive findings but potential hypotheses for future functional validation and clinical translation.
Table 7. Emergent biological knowledge from CyTOF/IMC analyses across cancers.
The Texstem-to-Texterm bifurcation, captured in CD8+ T cells via CyTOF and IMC in melanoma, NSCLC, and colorectal cancer, emerged as a pan-cancer axis of T-cell exhaustion. Several studies (53, 59, 67) demonstrated that PD-1int CD39− CD28+ Texstem CD8+ cells were enriched in responders to immune checkpoint inhibitors (ICI), whereas PD-1hi TIM-3+ CD39+ Texterm cells predominated in non-responders.
A myeloid-barrier archetype characterised by Arg1hi and CD38+ tumour-associated macrophages (TAMs), often co-localised with α-SMA+ CAFs, was described in urothelial, pancreatic, and hepatocellular carcinomas (68, 72, 80). These immunosuppressive cell clusters were spatially associated with T-cell exclusion zones and ICI resistance, suggesting potential utility of CD38/Arg1 as combinatorial therapeutic targets.
Tertiary lymphoid structures (TLSs) with mature CXCL13+CD21+BCL6+ cell aggregates were associated with favourable prognosis in HCC (69), HPV-positive oropharyngeal cancer (37), and Hodgkin lymphoma (41). TLS maturity scores derived from IMC were predictive of ICI benefit in some of these settings.
Taken together, these motifs underscore that CyTOF and IMC are not merely descriptive but hypothesis-generating technologies, revealing recurrent immune organisational structures that may inform biomarker development, patient stratification, and mechanistically guided therapeutic design.
4 Discussion
4.1 Mechanistic perspectives: immunometabolism, T-Cell plasticity, and cytokine dysregulation
Beyond descriptive phenotyping, high-dimensional cytometry offers a powerful window into the mechanistic underpinnings of tumour–immune interactions. Three emerging axes, immunometabolism, T-cell plasticity, and cytokine network dynamics, deserve greater integration into single-cell analyses to support functional interpretation and therapeutic targeting.
4.2 Immunometabolism and suppressive niches
Several CyTOF and IMC studies have uncovered metabolically defined immunosuppressive niches, particularly in pancreatic ductal adenocarcinoma and bladder cancer, where extracellular adenosine (eAdo) accumulates in hypoxic regions enriched with Arg1hi macrophages and FOXP3+ regulatory T cells (62, 78). Other panels incorporating transferrin receptor (TFRC), CD71, and enzymes of the adenosine pathway (CD39/CD73) have demonstrated that metabolic reprogramming shapes T-cell exhaustion and myeloid immunosuppression (23, 80, 89). These findings underscore the utility of integrating metabolic markers into CyTOF workflows to map immunometabolic circuits and identify actionable targets such as CD73, IDO1, or MCT1.
4.3 T-cell plasticity and lineage bifurcation
Recent high-dimensional profiling has illuminated dynamic transitions between progenitor-like (Texstem) and terminally exhausted (Texterm) CD8+ T-cell states, often coexisting within the same tumour (53, 54, 67). Moreover, cytokine-rich environments can drive unconventional phenotypes, such as CD4+ cytotoxic T lymphocytes or hybrid Treg–Th17 cells, which are now detectable via CyTOF when panels include granzyme/perforin and lineage transcription factors (e.g., FOXP3, RORγt, T-bet) (90). These data reinforce the concept that T-cell states represent a continuum modulated by antigenic stimulation, epigenetic imprinting, and cytokine exposure.
4.4 Cytokine and chemokine dysregulation
While surface phenotyping remains dominant, select studies have incorporated intracellular cytokine readouts (e.g., IL-10, TNF-α, IFN-γ), revealing differential cytokine bursts associated with response or resistance to immunotherapy (47, 59). Such profiles can be used to stratify immune “inflamed” versus “exhausted” microenvironments and may predict irAE risk. The integration of cytokine data into supervised learning models has already yielded predictive signatures in melanoma and NSCLC cohorts (53, 91).
Incorporating these mechanistic layers, metabolic profiling, plasticity trajectories, and cytokine output, into future CyTOF/IMC studies will deepen biological insight and enhance biomarker robustness across cancer types.
4.5 Clinical relevance and use in trial stratification
Although the review outlines the diversity of CyTOF and IMC applications, it is important to further emphasise their tangible impact on clinical practice and translational oncology. Several studies have gone beyond descriptive profiling and leveraged CyTOF- or IMC-derived signatures to guide patient stratification, therapeutic choices, or response monitoring in clinical trials.
In particular, CyTOF-defined T-cell states, such as progenitor exhausted (Texstem) CD8+ subsets, have been associated with durable responses to immune checkpoint blockade, influencing eligibility or stratification in melanoma, NSCLC, and colorectal cancer trials (53, 67, 70). Conversely, the enrichment of terminally exhausted PD-1hi TIM-3+ CD8+ T cells in non-responders has helped identify patients less likely to benefit from anti-PD-1 monotherapy (53, 57).
In the spatial context, IMC-derived metrics such as the α-SMA+CAF–CD163+ macrophage perimeter score in oesophageal squamous cell carcinoma (68) and CXCR5+–CXCL13+ spatial pairings in Hodgkin lymphoma (74) have stratified survival risk independently of classical histopathological parameters. These spatial biomarkers are now being considered as decision tools for tailoring follow-up or selecting intensified regimens.
Additionally, longitudinal CyTOF profiling has supported early prediction of treatment efficacy in vaccine-based immunotherapy trials, as seen in metastatic pancreatic cancer (45), and has informed in silico modelling of drug combinations based on spatial resistance niches in hepatocellular carcinoma (70).
4.6 Conserved archetypes of tumour–immune spatial organization
Our synthesis reveals that certain immune organizational structures, previously reported as context-specific, actually recur across multiple tumour types and analytical platforms. For instance, CD38+/Arg1+ myeloid barriers, initially described as immunosuppressive niches in pancreatic cancer and hepatocellular carcinoma, now emerge as recurrent spatial motifs that may impede CD8+ T-cell infiltration (92, 93). Similarly, tertiary lymphoid structures (TLS) with high maturity, marked by CXCL13+ dendritic cells and germinal-center-like B-cell organization, appeared in independent studies across HCC, Hodgkin lymphoma, and HPV-positive head and neck cancer, consistently correlating with enhanced immunotherapy response.
Likewise, on the functional axis of CD8+ T-cell exhaustion, a bifurcation between stem−like (Texstem/Tpex) and terminally exhausted (Texterm) states has emerged as a pan-cancer conserved phenomenon. Stem-like TCF−1+PD−1+ progenitor exhausted (Tpex) or Texstem cells are enriched in responders to checkpoint blockade and localize within TLS or tumour-draining lymph nodes (94–96). In contrast, PD−1hi TIM−3+ CD39+ Texterm cells, found in multiple tumour contexts, are associated with poor response and therapeutic resistance (94, 95, 97, 98). Taken together, our spatial meta-analysis delineates three conserved archetypes of tumour–immune organization, myeloid exclusionary barriers, TLS maturity, and CD8 T-cell bifurcation, that recur across cancer types and platforms. Their repeated observation suggests that they represent pan-cancer axes of immune–tumour interaction, whose shared regulatory mechanisms (e.g., metabolic constraints, stromal interactions, epigenetic programming) merit further investigation as potential universal biomarkers and therapeutic targets.
4.7 Methodological limitations and biases
Despite the richness of data extracted from CyTOF and IMC studies, several methodological limitations must be acknowledged when interpreting their findings collectively. A substantial proportion of the included studies were conducted in small cohorts, often ≤30 patients, limiting statistical power and the generalisability of identified immune signatures. For example, many exploratory studies reported associations between rare T-cell subsets or spatial arrangements and clinical outcomes without independent validation cohorts (54, 68, 75).
Moreover, external validation remains the exception rather than the rule. Only a few studies employed prospective replication or applied discovered signatures across multiple datasets, reducing the robustness of proposed biomarkers (45, 61). The field also remains vulnerable to publication bias, as studies demonstrating predictive or prognostic significance are more likely to be published, potentially inflating perceived reproducibility.
Finally, heterogeneity in panel composition, staining protocols, and analytical pipelines further complicates cross-study comparison. While recent efforts toward standardised reference panels and batch-correction frameworks are promising (99, 100), broader adoption is needed to support multi-centre reproducibility and meta-analytic synthesis.
Most solid-tumour CyTOF studies applied standard enzymatic dissociation protocols combining collagenase IV and DNase I, occasionally supplemented by hyaluronidase, to generate single-cell suspensions from fresh tumour specimens. These approaches, while widely used and reproducible, may introduce biases by preferentially preserving lymphoid and myeloid subsets over more fragile populations such as granulocytes or stromal cells.
5 Conclusion
Our pan-cancer synthesis of CyTOF/IMC studies reveals a consistent convergence of immune markers onto five core metabolic axes of the tumour microenvironment: (i) glycolysis (Warburg-like fuelling of activation and cytotoxicity), (ii) oxidative phosphorylation (OXPHOS) (mitochondrial programs underpinning memory and cellular longevity), (iii) fatty-acid oxidation/lipid metabolism (FAO) (predominant in Tregs and tissue-repair macrophages), (iv) amino-acid metabolism (arginine/tryptophan/glutamine and NAD+/adenosine pathways shaping exhaustion and myeloid polarization), and (v) nucleotide biosynthesis (proliferative demand in Ki-67+ lymphocytes). This unified framework clarifies the bioenergetic dependencies behind observed immune phenotypes and points to shared therapeutic vulnerabilities. Beyond this metabolic map, three recurrent spatial–immunological motifs emerge (1): the CD8+ Tex bifurcation between a TCF1+ PD-1int progenitor reservoir (Texstem) and a PD-1hi TIM-3+ terminal state (Texterm) that governs the durability of checkpoint responses (2); a CD38+/Arg1hi myeloid barrier that couples arginine depletion with NAD+ consumption to enforce T-cell exclusion and therapy resistance; and (3) the TLS maturity axis (CXCL13+ aggregates) in which germinal-center–like, mature TLS function as intratumoural priming hubs and associate with superior prognosis and immunotherapy benefit. Together, these metabolic axes and spatial motifs provide a common language to design composite spatial–metabolic signatures, standardize measurement panels, and guide rational interventions (e.g., CD38/Arg1 targeting, TLS maturation, exhaustion reprogramming) to be prospectively validated in harmonized, multicentre cohorts.
Author contributions
AV: Conceptualization, Data curation, Formal Analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. AD: Conceptualization, Data curation, Formal Analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. MA: Investigation, Resources, Funding acquisition, Conceptualization, Writing – review & editing, Methodology, Validation, Visualization, Formal Analysis, Supervision, Software, Data curation, Writing – original draft, Project administration.
Funding
The author(s) declare that no financial support was received for the research, and/or publication of this article.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declare that no Generative AI was used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2025.1672312/full#supplementary-material
References
1. Maleki EH, Bahrami AR, and Matin MM. Cancer cell cycle heterogeneity as a critical determinant of therapeutic resistance. Genes Dis. (2023) 11:189–204. doi: 10.1016/j.gendis.2022.11.025
2. McGranahan N and Swanton C. Clonal heterogeneity and tumour evolution: past, present, and the future. Cell. (2017) 168:613–28. doi: 10.1016/j.cell.2017.01.018
3. de Visser KE and Joyce JA. The evolving tumour microenvironment: From cancer initiation to metastatic outgrowth. Cancer Cell. (2023) 41:374–403. doi: 10.1016/j.ccell.2023.02.016
4. Gohil SH, Iorgulescu JB, Braun DA, Keskin DB, and Livak KJ. Applying high-dimensional single-cell technologies to the analysis of cancer immunotherapy. Nat Rev Clin Oncol. (2021) 18:244–56. doi: 10.1038/s41571-020-00449-x
5. Spitzer MH and Nolan GP. Mass cytometry: single cells, many features. Cell. (2016) 165:780–91. doi: 10.1016/j.cell.2016.04.019
6. Gonzalez VD, Huang Y-W, and Fantl WJ. Mass cytometry for the characterization of individual cell types in ovarian solid tumours. Methods Mol Biol Clifton NJ. (2022) 2424:59–94. doi: 10.1007/978-1-0716-1956-8_4
7. Atkuri KR, Stevens JC, and Neubert H. Mass cytometry: a highly multiplexed single-cell technology for advancing drug development. Drug Metab Dispos Biol Fate Chem. (2015) 43:227–33. doi: 10.1124/dmd.114.060798
8. Bandura DR, Baranov VI, Ornatsky OI, Antonov A, Kinach R, Lou X, et al. Mass cytometry: technique for real time single cell multitarget immunoassay based on inductively coupled plasma time-of-flight mass spectrometry. Anal Chem. (2009) 81:6813–22. doi: 10.1021/ac901049w
9. Tanner SD, Baranov VI, Ornatsky OI, Bandura DR, and George TC. An introduction to mass cytometry: fundamentals and applications. Cancer Immunol Immunother CII. (2013) 62:955–65. doi: 10.1007/s00262-013-1416-8
10. de Castro C G, del Hierro A G, H-Vázquez J, Cuesta-Sancho S, and Bernardo D. State-of-the-art cytometry in the search of novel biomarkers in digestive cancers. Front Oncol. (2024) 14:1407580. doi: 10.3389/fonc.2024.1407580
11. Hu Z, Tang A, Singh J, Bhattacharya S, and Butte AJ. A robust and interpretable end-to-end deep learning model for cytometry data. Proc Natl Acad Sci. (2020) 117:21373–80. doi: 10.1073/pnas.2003026117
12. Hu Z, Bhattacharya S, and Butte AJ. Application of machine learning for cytometry data. Front Immunol. (2022) 12:787574. doi: 10.3389/fimmu.2021.787574
13. Fuda F, Chen M, Chen W, and Cox A. Artificial intelligence in clinical multiparameter flow cytometry and mass cytometry–key tools and progress. Semin Diagn Pathol. (2023) 40:120–8. doi: 10.1053/j.semdp.2023.02.004
14. Schlecht A, Boneva S, Salie H, Killmer S, Wolf J, Hajdu RI, et al. Imaging mass cytometry for high-dimensional tissue profiling in the eye. BMC Ophthalmol. (2021) 21:338. doi: 10.1186/s12886-021-02099-8
15. Chang Q, Ornatsky OI, Siddiqui I, Loboda A, Baranov VI, and Hedley DW. Imaging mass cytometry. Cytometry A. (2017) 91:160–9. doi: 10.1002/cyto.a.23053
16. Giesen C, Wang HAO, Schapiro D, Zivanovic N, Jacobs A, Hattendorf B, et al. Highly multiplexed imaging of tumour tissues with subcellular resolution by mass cytometry. Nat Methods. (2014) 11:417–22. doi: 10.1038/nmeth.2869
17. Seliger B and Massa C. CyTOF as a suitable tool for stratification and monitoring of cancer patients. J Transl Med. (2025) 23:724. doi: 10.1186/s12967-025-06764-0
18. Ma L-L, Wang Y-Y, Yang Z-H, Huang D, Weng H, and Zeng X-T. Methodological quality (risk of bias) assessment tools for primary and secondary medical studies: what are they and which is better? Mil Med Res. (2020) 7:7. doi: 10.1186/s40779-020-00238-8
19. Sterne JA, Hernán MA, Reeves BC, Savović J, Berkman ND, Viswanathan M, et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ. (2016) 355:i4919. doi: 10.1136/bmj.i4919
20. Whiting PF, Rutjes AWS, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2 Group. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. (2011) 155:529–36. doi: 10.7326/0003-4819-155-8-201110180-00009
21. Jiang X, Wang J, Wang M, Xuan M, Han S, Li C, et al. ITGB4 as a novel serum diagnosis biomarker and potential therapeutic target for colorectal cancer. Cancer Med. (2021) 10:6823–34. doi: 10.1002/cam4.4216
22. Wang M, Huang Y-K, Kong JC, Sun Y, Tantalo DG, Yeang HXA, et al. High-dimensional analyses reveal a distinct role of T-cell subsets in the immune microenvironment of gastric cancer. Clin Transl Immunol. (2020) 9:e1127. doi: 10.1002/cti2.1127
23. Song F, Hu B, Liang X-L, Cheng J-W, Wang C-G, Wang P-X, et al. Anlotinib potentiates anti-PD1 immunotherapy via transferrin receptor-dependent CD8+ T-cell infiltration in hepatocellular carcinoma. Clin Transl Med. (2024) 14:e1738. doi: 10.1002/ctm2.1738
24. Nissen MD, Kusakabe M, Wang X, Simkin G, Gracias D, Tyshchenko K, et al. Single cell phenotypic profiling of 27 DLBCL cases reveals marked intertumoural and intratumoural heterogeneity. Cytom Part J Int Soc Anal Cytol. (2020) 97:620–9. doi: 10.1002/cyto.a.23919
25. Figueiredo CR, Kalirai H, Sacco JJ, Azevedo RA, Duckworth A, Slupsky JR, et al. Loss of BAP1 expression is associated with an immunosuppressive microenvironment in uveal melanoma, with implications for immunotherapy development. J Pathol. (2020) 250:420–39. doi: 10.1002/path.5384
26. Panovska D, Nazari P, Cole B, Creemers P-J, Derweduwe M, Solie L, et al. Single-cell molecular profiling using ex vivo functional readouts fuels precision oncology in glioblastoma. Cell Mol Life Sci CMLS. (2023) 80:147. doi: 10.1007/s00018-023-04772-1
27. Qing M, Zhou T, Perova T, Abraham Y, Sweeney C, Krevvata M, et al. Immune profiling of patients with extranodal natural killer/T cell lymphoma treated with daratumumab. Ann Hematol. (2024) 103:1989–2001. doi: 10.1007/s00277-023-05603-w
28. Gide TN, Quek C, Menzies AM, Tasker AT, Shang P, Holst J, et al. Distinct immune cell populations define response to anti-PD-1 monotherapy and anti-PD-1/anti-CTLA-4 combined therapy. Cancer Cell. (2019) 35:238–255.e6. doi: 10.1016/j.ccell.2019.01.003
29. De Vargas Roditi L, Jacobs A, Rueschoff JH, Bankhead P, Chevrier S, Jackson HW, et al. Single-cell proteomics defines the cellular heterogeneity of localized prostate cancer. Cell Rep Med. (2022) 3:100604. doi: 10.1016/j.xcrm.2022.100604
30. Mer AS, Heath EM, Madani Tonekaboni SA, Dogan-Artun N, Nair SK, Murison A, et al. Biological and therapeutic implications of a unique subtype of NPM1 mutated AML. Nat Commun. (2021) 12:1054. doi: 10.1038/s41467-021-21233-0
31. Chauhan SK, Dunn C, Andresen NK, Røssevold AH, Skorstad G, Sike A, et al. Peripheral immune cells in metastatic breast cancer patients display a systemic immunosuppressed signature consistent with chronic inflammation. NPJ Breast Cancer. (2024) 10:30. doi: 10.1038/s41523-024-00638-2
32. Sun W, Qiu F, Zheng J, Fang L, Qu J, Zhang S, et al. CD57-positive CD8 + T cells define the response to anti-programmed cell death protein-1 immunotherapy in patients with advanced non-small cell lung cancer. NPJ Precis Oncol. (2024) 8:25. doi: 10.1038/s41698-024-00513-0
33. Goldberg L, Haas ER, Vyas V, Urak R, Forman SJ, and Wang X. Single-cell analysis by mass cytometry reveals CD19 CAR T cell spatiotemporal plasticity in patients. Oncoimmunology. (2022) 11:2040772. doi: 10.1080/2162402X.2022.2040772
34. Lu Y, Xue G, Zheng N, Han K, Yang W, Wang R-S, et al. hDirect-MAP: projection-free single-cell modeling of response to checkpoint immunotherapy. Brief Bioinform. (2022) 23:bbab575. doi: 10.1093/bib/bbab575
35. Raju Paul S, Valiev I, Korek SE, Zyrin V, Shamsutdinova D, Gancharova O, et al. B cell-dependent subtypes and treatment-based immune correlates to survival in stage 3 and 4 lung adenocarcinomas. FASEB BioAdv. (2023) 5:156–70. doi: 10.1096/fba.2023-00009
36. Smelser WW, Wang J, Ogden KM, Chang SS, and Kirschner AN. Intravesical oncolytic virotherapy and immunotherapy for non-muscle-invasive bladder cancer mouse model. BJU Int. (2023) 132:298–306. doi: 10.1111/bju.16012
37. Rodriguez E, Zwart ES, Affandi AA, Verhoeff J, de Kok M, Boyd LNC, et al. In-depth immune profiling of peripheral blood mononuclear cells in patients with pancreatic ductal adenocarcinoma reveals discriminative immune subpopulations. Cancer Sci. (2024) 115:2170–83. doi: 10.1111/cas.16147
38. Zhou L, Zeng Z, Egloff AM, Zhang F, Guo F, Campbell KM, et al. Checkpoint blockade-induced CD8+ T cell differentiation in head and neck cancer responders. J Immunother Cancer. (2022) 10:e004034. doi: 10.1136/jitc-2021-004034
39. Lheureux S, Matei DE, Konstantinopoulos PA, Wang BX, Gadalla R, Block MS, et al. Translational randomized phase II trial of cabozantinib in combination with nivolumab in advanced, recurrent, or metastatic endometrial cancer. J Immunother Cancer. (2022) 10:e004233. doi: 10.1136/jitc-2021-004233
40. Krieg C, Weber LM, Fosso B, Marzano M, Hardiman G, Olcina MM, et al. Complement downregulation promotes an inflammatory signature that renders colorectal cancer susceptible to immunotherapy. J Immunother Cancer. (2022) 10:e004717. doi: 10.1136/jitc-2022-004717
41. Fehlings M, Kim L, Guan X, Yuen K, Tafazzol A, Sanjabi S, et al. Single-cell analysis reveals clonally expanded tumour-associated CD57+ CD8 T cells are enriched in the periphery of patients with metastatic urothelial cancer responding to PD-L1 blockade. J Immunother Cancer. (2022) 10:e004759. doi: 10.1136/jitc-2022-004759
42. Edwards J, Tasker A, Pires da Silva I, Quek C, Batten M, Ferguson A, et al. Prevalence and cellular distribution of novel immune checkpoint targets across longitudinal specimens in treatment-naïve melanoma patients: implications for clinical trials. Clin Cancer Res. (2019) 25:3247–58. doi: 10.1158/1078-0432.CCR-18-4011
43. Li B, Li G, Yan X, Zhu D, Lin PP, Wang Z, et al. Fresh tissue multi-omics profiling reveals immune classification and suggests immunotherapy candidates for conventional chondrosarcoma. Clin Cancer Res. (2021) 27:6543–58. doi: 10.1158/1078-0432.CCR-21-1893
44. Okuma HS, Watanabe K, Tsuchihashi K, Machida R, Sadachi R, Hirakawa A, et al. Phase II trial of nivolumab in metastatic rare cancer with dMMR or MSI-H and relation with immune phenotypic analysis (the ROCK trial). Clin Cancer Res. (2023) 29:5079–86. doi: 10.1158/1078-0432.CCR-23-1807
45. Nair N, Chen S-Y, Lemmens E, Chang S, Le DT, Jaffee EM, et al. Single-cell immune competency signatures associate with survival in phase II GVAX and CRS-207 randomized studies in patients with metastatic pancreatic cancer. Cancer Immunol Res. (2020) 8:609–17. doi: 10.1158/2326-6066.CIR-19-0650
46. Ho WJ, Yarchoan M, Charmsaz S, Munday RM, Danilova L, Sztein MB, et al. Multipanel mass cytometry reveals anti-PD-1 therapy-mediated B and T cell compartment remodeling in tumour-draining lymph nodes. JCI Insight. (2020) 5:e132286. doi: 10.1172/jci.insight.132286
47. Luo Y-H, Shen C-I, Chiang C-L, and Chen Y-M. Immune signatures of patients with advanced non-small-cell lung cancer for efficacy prediction after immunotherapy. Ther Adv Med Oncol. (2024) 16:17588359241284946. doi: 10.1177/17588359241284946
48. Duault C, Kumar A, Taghi Khani A, Lee SJ, Yang L, Huang M, et al. Activated natural killer cells predict poor clinical prognosis in high-risk B- and T-cell acute lymphoblastic leukaemia. Blood. (2021) 138:1465–80. doi: 10.1182/blood.2020009871
49. Zhang Q, Zhang W, Lin T, Lu W, He X, Ding Y, et al. Mass cytometry reveals immune atlas of urothelial carcinoma. BMC Cancer. (2022) 22:677. doi: 10.1186/s12885-022-09788-7
50. Dong Y, Wang Z, Xie G-F, Li C, Zuo W-W, Meng G, et al. Pregnane X receptor is associated with unfavorable survival and induces chemotherapeutic resistance by transcriptional activating multidrug resistance-related protein 3 in colorectal cancer. Mol Cancer. (2017) 16:71. doi: 10.1186/s12943-017-0641-8
51. Liu Z, Li X, Gao Y, Liu J, Feng Y, Liu Y, et al. Epigenetic reprogramming of Runx3 reinforces CD8 + T-cell function and improves the clinical response to immunotherapy. Mol Cancer. (2023) 22:84. doi: 10.1186/s12943-023-01768-0
52. Davis-Marcisak EF, Fitzgerald AA, Kessler MD, Danilova L, Jaffee EM, Zaidi N, et al. Transfer learning between preclinical models and human tumours identifies a conserved NK cell activation signature in anti-CTLA-4 responsive tumours. Genome Med. (2021) 13:129. doi: 10.1186/s13073-021-00944-5
53. Subrahmanyam PB, Dong Z, Gusenleitner D, Giobbie-Hurder A, Severgnini M, Zhou J, et al. Distinct predictive biomarker candidates for response to anti-CTLA-4 and anti-PD-1 immunotherapy in melanoma patients. J Immunother Cancer. (2018) 6:18. doi: 10.1186/s40425-018-0328-8
54. Ye L, Ryu H, Granadier D, Nguyen LT, Simoni Y, Dick I, et al. Stem-like exhausted CD8 T cells in pleural effusions predict improved survival in non-small cell lung cancer (NSCLC) and mesothelioma. Transl Lung Cancer Res. (2024) 13:2352–72. doi: 10.21037/tlcr-24-284
55. Ferrant J, Lhomme F, Le Gallou S, Irish JM, and Roussel M. Circulating myeloid regulatory cells: promising biomarkers in B-cell lymphomas. Front Immunol. (2020) 11:623993. doi: 10.3389/fimmu.2020.623993
56. Le Gallou S, Lhomme F, Irish JM, Mingam A, Pangault C, Monvoisin C, et al. Nonclassical monocytes are prone to migrate into tumour in diffuse large B-cell lymphoma. Front Immunol. (2021) 12:755623. doi: 10.3389/fimmu.2021.755623
57. Luo Y, Zong Y, Hua H, Gong M, Peng Q, Li C, et al. Immune-infiltrating signature-based classification reveals CD103+CD39+ T cells associate with colorectal cancer prognosis and response to immunotherapy. Front Immunol. (2022) 13:1011590. doi: 10.3389/fimmu.2022.1011590
58. Govindarajan A, Salgia NJ, Li H, Castro DV, Mirzapoiazova T, Armstrong B, et al. Characterization of papillary and clear cell renal cell carcinoma through imaging mass cytometry reveals distinct immunologic profiles. Front Immunol. (2023) 14:1182581. doi: 10.3389/fimmu.2023.1182581
59. Luo Y-H, Shen C-I, Chiang C-L, Huang H-C, and Chen Y-M. Dynamic immune signatures of patients with advanced non-small-cell lung cancer for infection prediction after immunotherapy. Front Immunol. (2024) 15:1269253. doi: 10.3389/fimmu.2024.1269253
60. Goshen-Lago T, Szwarcwort-Cohen M, Benguigui M, Almog R, Turgeman I, Zaltzman N, et al. The potential role of immune alteration in the cancer-COVID19 equation-A prospective longitudinal study. Cancers. (2020) 12:2421. doi: 10.3390/cancers12092421
61. Thomsen LCV, Kleinmanns K, Anandan S, Gullaksen S-E, Abdelaal T, Iversen GA, et al. Combining mass cytometry data by CyTOFmerge reveals additional cell phenotypes in the heterogeneous ovarian cancer tumour microenvironment: A pilot study. Cancers. (2023) 15:5106. doi: 10.3390/cancers15205106
62. Lu Q, Wang J, Tao Y, Zhong J, Zhang Z, Feng C, et al. Small cajal body-specific RNA12 promotes carcinogenesis through modulating extracellular matrix signaling in bladder cancer. Cancers. (2024) 16:483. doi: 10.3390/cancers16030483
63. Yoshida Y, Yoshio S, Yamazoe T, Mori T, Tsustui Y, Kawai H, et al. Phenotypic characterization by single-cell mass cytometry of human intrahepatic and peripheral NK cells in patients with hepatocellular carcinoma. Cells. (2021) 10:1495. doi: 10.3390/cells10061495
64. Lee H-S, Jang H-J, Choi JM, Zhang J, de Rosen VL, Wheeler TM, et al. Comprehensive immunoproteogenomic analyses of Malignant pleural mesothelioma. JCI Insight. (2018) 3:e98575. doi: 10.1172/jci.insight.98575
65. Pizzolla A, Keam SP, Vergara IA, Caramia F, Thio N, Wang M, et al. Tissue-resident memory T cells from a metastatic vaginal melanoma patient are tumour-responsive T cells and increase after anti-PD-1 treatment. J Immunother Cancer. (2022) 10:e004574. doi: 10.1136/jitc-2022-004574
66. Lu Q, Gao J, Tang S, Li Z, Wang X, Deng C, et al. Integrated RNA sequencing and single-cell mass cytometry reveal a novel role of lncRNA HOXA-AS2 in tumourigenesis and stemness of hepatocellular carcinoma. OncoTargets Ther. (2020) 13:10901–16. doi: 10.2147/OTT.S272717
67. Wang XQ, Danenberg E, Huang C-S, Egle D, Callari M, Bermejo B, et al. Spatial predictors of immunotherapy response in triple-negative breast cancer. Nature. (2023) 621:868–76. doi: 10.1038/s41586-023-06498-3
68. He J-Z, Chen Y, Zeng F-M, Huang Q-F, Zhang H-F, Wang S-H, et al. Spatial analysis of stromal signatures identifies invasive front carcinoma-associated fibroblasts as suppressors of anti-tumour immune response in esophageal cancer. J Exp Clin Cancer Res CR. (2023) 42:136. doi: 10.1186/s13046-023-02697-y
69. Mi H, Ho WJ, Yarchoan M, and Popel AS. Multi-scale spatial analysis of the tumour microenvironment reveals features of cabozantinib and nivolumab efficacy in hepatocellular carcinoma. Front Immunol. (2022) 13:892250. doi: 10.3389/fimmu.2022.892250
70. Zhang S, Deshpande A, Verma BK, Wang H, Mi H, Yuan L, et al. Integration of clinical trial spatial multiomics analysis and virtual clinical trials enables immunotherapy response prediction and biomarker discovery. Cancer Res. (2024) 84:2734–48. doi: 10.1158/0008-5472.CAN-24-0943
71. Gan X, Dong W, You W, Ding D, Yang Y, Sun D, et al. Spatial multimodal analysis revealed tertiary lymphoid structures as a risk stratification indicator in combined hepatocellular-cholangiocarcinoma. Cancer Lett. (2024) 581:216513. doi: 10.1016/j.canlet.2023.216513
72. Xiang H, Pan Y, Sze MA, Wlodarska M, Li L, van de Mark KA, et al. Single-cell analysis identifies NOTCH3-mediated interactions between stromal cells that promote microenvironment remodeling and invasion in lung adenocarcinoma. Cancer Res. (2024) 84:1410–25. doi: 10.1158/0008-5472.CAN-23-1183
73. Rayford A, Gärtner F, Ramnefjell MP, Lorens JB, Micklem DR, Aanerud M, et al. AXL expression reflects tumour-immune cell dynamics impacting outcome in non-small cell lung cancer patients treated with immune checkpoint inhibitor monotherapy. Front Immunol. (2024) 15:1444007. doi: 10.3389/fimmu.2024.1444007
74. Aoki T, Jiang A, Xu A, Yin Y, Gamboa A, Milne K, et al. Spatially resolved tumour microenvironment predicts treatment outcomes in relapsed/refractory hodgkin lymphoma. J Clin Oncol. (2024) 42:1077–87. doi: 10.1200/JCO.23.01115
75. Martinez-Morilla S, Villarroel-Espindola F, Wong PF, Toki MI, Aung TN, Pelekanou V, et al. Biomarker discovery in patients with immunotherapy-treated melanoma with imaging mass cytometry. Clin Cancer Res. (2021) 27:1987–96. doi: 10.1158/1078-0432.CCR-20-3340
76. Xiao X, Guo Q, Cui C, Lin Y, Zhang L, Ding X, et al. Multiplexed imaging mass cytometry reveals distinct tumour-immune microenvironments linked to immunotherapy responses in melanoma. Commun Med. (2022) 2:131. doi: 10.1038/s43856-022-00197-2
77. Abdulrahman Z, Santegoets SJ, Sturm G, Charoentong P, Ijsselsteijn ME, Somarakis A, et al. Tumour-specific T cells support chemokine-driven spatial organization of intratumoural immune microaggregates needed for long survival. J Immunother Cancer. (2022) 10:e004346. doi: 10.1136/jitc-2021-004346
78. Graziano V, Dannhorn A, Hulme H, Williamson K, Buckley H, Karim SA, et al. Defining the spatial distribution of extracellular adenosine revealed a myeloid-dependent immunosuppressive microenvironment in pancreatic ductal adenocarcinoma. J Immunother Cancer. (2023) 11:e006457. doi: 10.1136/jitc-2022-006457
79. Tao Y, Li X, Zhang Y, He L, Lu Q, Wang Y, et al. TP53-related signature for predicting prognosis and tumour microenvironment characteristics in bladder cancer: A multi-omics study. Front Genet. (2022) 13:1057302. doi: 10.3389/fgene.2022.1057302
80. Wang Y, Song W, Feng C, Wu S, Qin Z, Liu T, et al. Multi-omics analysis unveils the predictive value of IGF2BP3/SPHK1 signaling in cancer stem cells for prognosis and immunotherapeutic response in muscle-invasive bladder cancer. J Transl Med. (2024) 22:900. doi: 10.1186/s12967-024-05685-8
81. Barsch M, Salié H, Schlaak AE, Zhang Z, Hess M, Mayer LS, et al. T-cell exhaustion and residency dynamics inform clinical outcomes in hepatocellular carcinoma. J Hepatol. (2022) 77:397–409. doi: 10.1016/j.jhep.2022.02.032
82. Liu S, Liao S, Liang L, Deng J, and Zhou Y. The relationship between CD4+ T cell glycolysis and their functions. Trends Endocrinol Metab TEM. (2023) 34:345–60. doi: 10.1016/j.tem.2023.03.006
83. Vallée A and Lecarpentier Y. Crosstalk between peroxisome proliferator-activated receptor gamma and the canonical WNT/β-catenin pathway in chronic inflammation and oxidative stress during carcinogenesis. Front Immunol. (2018) 9:745. doi: 10.3389/fimmu.2018.00745
84. Yan Y, Huang L, Liu Y, Yi M, Chu Q, Jiao D, et al. Metabolic profiles of regulatory T cells and their adaptations to the tumour microenvironment: implications for antitumour immunity. J Hematol OncolJ Hematol Oncol. (2022) 15:104. doi: 10.1186/s13045-022-01322-3
85. Zhang S, Lv K, Liu Z, Zhao R, and Li F. Fatty acid metabolism of immune cells: a new target of tumour immunotherapy. Cell Death Discov. (2024) 10:39. doi: 10.1038/s41420-024-01807-9
86. Bai D, Zhou Y, Jing L, Guo C, and Yang Q. Arginine metabolism in cancer biology and immunotherapy. Immune Netw. (2025) 25:e30. doi: 10.4110/in.2025.25.e30
87. O’Sullivan D and Pearce EL. Targeting T cell metabolism for therapy. Trends Immunol. (2015) 36:71–80. doi: 10.1016/j.it.2014.12.004
88. Sun X and Kaufman PD. Ki-67: more than a proliferation marker. Chromosoma. (2018) 127:175–86. doi: 10.1007/s00412-018-0659-8
89. Nunes JB, Ijsselsteijn ME, Abdelaal T, Ursem R, van der Ploeg M, Giera M, et al. Integration of mass cytometry and mass spectrometry imaging for spatially resolved single-cell metabolic profiling. Nat Methods. (2024) 21:1796–800. doi: 10.1038/s41592-024-02392-6
90. Wlosik J, Granjeaud S, Gorvel L, Olive D, and Chretien A-S. A beginner’s guide to supervised analysis for mass cytometry data in cancer biology. Cytom Part J Int Soc Anal Cytol. (2024) 105:853–69. doi: 10.1002/cyto.a.24901
91. Bollhagen A and Bodenmiller B. Highly multiplexed tissue imaging in precision oncology and translational cancer research. Cancer Discov. (2024) 14:2071–88. doi: 10.1158/2159-8290.CD-23-1165
92. Ferkel SAM, Holman EA, Sojwal RS, Rubin SJS, and Rogalla S. Tumour-infiltrating immune cells in colorectal cancer. Neoplasia N Y N. (2024) 59:101091. doi: 10.1016/j.neo.2024.101091
93. Zander R and Cui W. Exhausted CD8+ T cells face a developmental fork in the road. Trends Immunol. (2023) 44:276–86. doi: 10.1016/j.it.2023.02.006
94. Ni L. Potential mechanisms of cancer stem-like progenitor T-cell bio-behaviours. Clin Transl Med. (2024) 14:e1817. doi: 10.1002/ctm2.1817
95. Wang D, Fang J, Wen S, Li Q, Wang J, Yang L, et al. A comprehensive profile of TCF1+ progenitor and TCF1– terminally exhausted PD-1+CD8+ T cells in head and neck squamous cell carcinoma: implications for prognosis and immunotherapy. Int J Oral Sci. (2022) 14:8. doi: 10.1038/s41368-022-00160-w
96. Zehn D, Thimme R, Lugli E, de Almeida GP, and Oxenius A. ‘Stem-like’ precursors are the fount to sustain persistent CD8+ T cell responses. Nat Immunol. (2022) 23:836–47. doi: 10.1038/s41590-022-01219-w
97. Wen S, Lu H, Wang D, Guo J, Dai W, and Wang Z. TCF-1 maintains CD8+ T cell stemness in tumour microenvironment. J Leukoc Biol. (2021) 110:585–90. doi: 10.1002/JLB.5MR1120-778R
98. Seyhan D, Allaire M, Fu Y, Conti F, Wang XW, Gao B, et al. Immune microenvironment in hepatocellular carcinoma: from pathogenesis to immunotherapy. Cell Mol Immunol. (2025) 22:1132–58. doi: 10.1038/s41423-025-01308-4
99. Hartmann FJ, Babdor J, Gherardini PF, Amir E-AD, Jones K, Sahaf B, et al. Comprehensive immune monitoring of clinical trials to advance human immunotherapy. Cell Rep. (2019) 28:819–831.e4. doi: 10.1016/j.celrep.2019.06.049
Keywords: CyTOF (mass cytometry), imaging mass cytometry, tumour microenvironment, single-cell proteomics, immune biomarkers, precision oncology, immune profiling
Citation: Vallée A, Drezet A and Arutkin M (2025) Pan-cancer convergence of tumour–immune microenvironment motifs revealed by CyTOF and imaging mass cytometry. Front. Immunol. 16:1672312. doi: 10.3389/fimmu.2025.1672312
Received: 24 July 2025; Accepted: 17 September 2025;
Published: 06 October 2025.
Edited by:
Rao Prabhala, Harvard Medical School, United StatesReviewed by:
Weian Zhu, Sun Yat-sen University, ChinaSrikanth Talluri, Dana–Farber Cancer Institute, United States
Lakshmi Bhavani Potluri, New York University, United States
Copyright © 2025 Vallée, Drezet and Arutkin. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Alexandre Vallée, YWwudmFsbGVlQGhvcGl0YWwtZm9jaC5jb20=