Skip Navigation
Skip to contents

Cancer Res Treat : Cancer Research and Treatment

Search
Advanced Search
OPEN ACCESS
mobile search button mobile menu button

Articles

Page Path
HOME > J Korean Cancer Assoc > Ahead-of print articles > Article
Original Article Immunosuppressive Tumor Microenvironment in Colorectal Cancer Lung Metastases: Implications for Recurrence after Metastasectomy
Minsuk Kwon1orcid, Min-Kyue Shin1,2orcid, Minae An3orcid, Yeong Jeong Jeon4orcid, Tae Hee Hong5, Jung Kyong Shin6, Sung Hee Lim1, Yoonah Park6, Yong Beom Cho6, Seung Tae Kim1, Yong Soo Choi4orcid, Jeeyun Lee1orcid

DOI: https://doi.org/10.4143/crt.2025.691
Published online: December 17, 2025

1Division of Hematology-Oncology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea

2Department of Digital Health, Samsung Advanced Institute of Health Science and Technology, Sungkyunkwan University School of Medicine, Seoul, Korea

3Samsung Precision Genome Medicine Institute, Samsung Medical Center, Seoul, Korea

4Department of Thoracic and Cardiovascular Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea

5Department of Thoracic Surgery, Yonsei University School of Medicine, Seoul, Korea

6Department of General Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea

Correspondence: Yong Soo Choi, Department of Thoracic and Cardiovascular Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 06351, Korea
Tel: 82-2-3410-6542 E-mail: ysooyah.choi@samsung.com
Co-correspondence: Jeeyun Lee, Division of Hematology-Oncology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 06351, Korea
Tel: 82-2-3410-3459 E-mail: jyunlee@skku.edu
*Minsuk Kwon, Min-Kyue Shin, Minae An, and Yeong Jeong Jeon contributed equally to this work.
• Received: July 3, 2025   • Accepted: December 8, 2025

Copyright © 2026 by the Korean Cancer Association

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

  • 1,631 Views
  • 128 Download
  • 1 Crossref
Full Article

Download PDF Download PDF

  • Purpose
    Colorectal cancer (CRC) lung metastases exhibit high recurrence rates after resection, underscoring the need for improved therapeutic strategies. This study aimed to characterize the tumor microenvironment of CRC lung metastases and identify the factors associated with recurrence.
  • Materials and Methods
    Fifteen CRC patients who underwent lung metastasectomy were enrolled. Multiplex immunohistochemistry (IHC), whole exome sequencing, transcriptome profiling, and single-cell RNA sequencing (scRNA-seq) were conducted on matched tumor, adjacent and distant normal lung tissues. Immune cell populations and gene expression profiles were analyzed and correlated with clinical recurrence outcomes.
  • Results
    Exome and transcriptome analyses revealed frequent TP53, KRAS, and APC mutations. Most tumors corresponded to consensus molecular subtypes 2 and 4, characterized by immune-depleted and fibrotic features. Tumors showed downregulation of effector T and natural killer (NK) cell signatures. IHC revealed reduced density and increased distance of CD8+ T cells and macrophages from the epithelial cells. scRNA-seq demonstrated increased regulatory T cells and decreased NK and effector T cells in tumor. Tumor-associated macrophages (TAMs), particularly SPP1 (osteopontin)-expressing subsets, were markedly enriched in tumor and correlated with suppressed effector T-cell activity. High SPP1 expression was associated with early recurrence and poor overall survival. Patients with recurrence had higher proportion of PD-1+ CD8+ T cells in adjacent normal tissues.
  • Conclusion
    Immunosuppressive features including enrichment of SPP1+ TAMs and depletion of effector T and NK cells contribute to recurrence after CRC lung metastasectomy. Therapeutic strategies targeting both TAMs and T cells may enhance clinical outcomes in this patient population.
  • Key words: Colorectal neoplasms, Lung metastasis, Recurrence, Tumor microenvironment, Mutation, Tumor-associated macrophages, T-lymphocytes
Colorectal cancer (CRC) is the second most common cause of cancer-related mortality globally, with most patients suffering from metastatic disease and surviving for less than three years. Improved clinical outcomes in metastatic CRC (mCRC) have been observed with the introduction of metastasectomy combined with perioperative systemic chemotherapy as well as local treatments, such as radiotherapy and ablation therapies [1]. For patients with CRC with resectable or potentially resectable liver metastases, curative resection combined with perioperative systemic chemotherapy has demonstrated a 5-year survival rate of 20%-45%. Similarly, resection of lung metastases has shown 5-year survival rates of 25%-35% in carefully selected patients [2], and practice guidelines recommend combined lung and liver metastasectomies for mCRC in select cases [3]. However, CRC patients undergoing extrahepatic resection have higher recurrence rates than those with liver-only resections, indicating the need for more effective systemic therapy following lung metastasectomy [4].
The current standard of initial therapy for mCRC includes cytotoxic chemotherapy with or without anti–vascular endothelial growth factor therapy, anti–epidermal growth factor receptor agents for patients with RAS/RAF wild-type tumors, and immune checkpoint inhibitors (ICIs) for patients with DNA mismatch repair deficiency (dMMR) or high microsatellite instability (MSI-H) [5,6]. Targeted agents have also shown clinical benefits in patients with mCRC with BRAF V600E mutations or HER-2 alterations [7,8]. Although ICIs provide substantial clinical advantages to a subset of patients with mCRC with dMMR/MSI-H tumors [9], their efficacy is limited to mismatch repair proficiency or microsatellite stable (MSS) mCRC, with objective response rates close to 0% in most studies [10]. Some combination therapies, such as nivolumab plus regorafenib, have shown promise in patients with MSS; however, further studies are necessary [11].
In patients treated with ICIs and at multiple sites of the disease, heterogeneous responses are common, with patients exhibiting varying results depending on the location of the metastases. Moreover, there is a trend toward differing responses to ICIs based on the metastatic organ involved [12]. The presence of liver [13] or bone metastases [14] has been associated with reduced survival rates. Conversely, lung metastases have been associated with favorable responses to anti–programmed death-1 (PD-1)–based immunotherapy [15]. The mechanisms of resistance to ICIs have been extensively studied, revealing that differences in tumor microenvironment (TME) properties between organs are significant explanatory factors [12]. Organ-specific resistance mechanisms may be attributed to variations in immune cell composition, the extracellular matrix, metabolic conditions, and the degree of vascular abnormality [16].
In this study, we aimed to investigate the characteristics of immune surveillance in tumors and adjacent normal tissues by analyzing the TME using whole-transcriptome sequencing (WTS) and evaluating the immune cell components using single-cell RNA sequencing (scRNA-seq) and multiplex immunohistochemistry (IHC).
We identified SPP1 (osteopontin)-expressing tumor-associated macrophages (TAMs) as key mediators of the immunosuppressive TME and examined their spatial and transcriptional association with T-cell dysfunction and clinical recurrence. To this end, we compared metastatic tumor tissues with adjacent and distant normal lung tissues to characterize immune alterations and assess prognostic associations based on tumor recurrence.
1. Sequencing of the whole exome and transcriptome
Genomic DNA (gDNA) was extracted from tumor tissues and matched blood samples using the AllPrep DNA/RNA Mini Kit (cat No. 80204, QIAGEN). We employed the Agilent SureSelect Target Enrichment procedure (Agilent Technologies) to prepare an Illumina paired-end sequencing library, starting with 1 μg of input gDNA to generate standard exome capture libraries. The SureSelect Human All Exon V6 Probe (Agilent Technologies) was used for each sample. The quantity and quality of DNA were assessed using agarose gel electrophoresis and PicoGreen. Following the manufacturer’s instructions, we diluted 1 μg of gDNA in elution buffer and sheared it via sonication using the Covaris LE220 focused ultrasonication device (Covaris Inc.) to achieve a desired peak size of 150-200 bp. The fragmented DNA was then repaired, and an “A” base was ligated to the 3′-end. Next, we amplified the fragments by polymerase chain reaction (PCR) after ligating the Agilent adapters. The resulting libraries were quantified using a TapeStation DNA Screen Tape D1000 (Agilent Technologies). According to the standard Agilent SureSelect Target Enrichment methodology, 250 ng of the DNA library was combined with hybridization buffer, blocking mixes, and 5 μg of RNase block from the SureSelect All Exon V6 capture library for exome capture. Hybridization of the capture baits was performed at 65°C for 24 hours using a heated thermal cycler lid at 105°C. The captured DNA was purified and amplified. The final purified product was quantified by quantitative PCR following the KAPA Library Quantification Protocol Guide for Illumina Sequencing platforms and verified using a TapeStation DNA ScreenTape D1000 (Agilent Technologies). Flow cell clusters were generated by multiplexing the samples using the TruSeq Rapid Cluster and TruSeq Rapid SBS kits (Illumina). The Illumina HiSeq2500 (Illumina) was used for paired-end (2×100 bp) sequencing of the indexed libraries.
2. Sequencing of single-cell transcriptome
Tumor and normal tissues were separated using gentleMACS, according to the manufacturer’s instructions for single-cell preparation. The cells were then cryopreserved in liquid nitrogen until further use. After thawing, the cell viability of each sample was approximately 90%. For each sample, a 10× Genomics cartridge containing up to 8,000 cells was used. Cell-barcoded 5′ gene expression libraries were created and sequenced at a depth of approximately 50,000 reads per cell using the Illumina NovaSeq 6000 platform. CellRanger ver. 5.0 (10× Genomics) was used to map libraries to the GRCh38 human reference genome.
3. Whole exome sequencing analysis
Whole exome sequencing (WES) reads were aligned to the reference human genome, GRCh37, using BWA-MEM [17]. The preprocessing steps were conducted using the Genome Analysis Toolkit (GATK, ver. 4.1.1.02), resulting in BAM files ready for analysis. The steps included duplicate marking, indel realignment, and base recalibration. We employed union variant calls from two tools, MuTect2 [18] and Strelka2 [19], to enhance the sensitivity in detecting somatic variations across both lower and higher allele frequencies in the provided tumor and paired normal BAM files. Both variant callers were run using the dbSNP ver. 138 for known polymorphic locations with default parameters. The Ensembl Variant Effect Predictor (VEP, release ver. 87) was used to annotate filtered variants with a minimum depth of ≥ 5 and minimum alternative alleles of ≥ 2 using the GRCh37 database.
4. WTS analysis
The RNA sequence reads were annotated using ENSEMBL ver. 98 and aligned to the human reference genome (GRCh38) using STAR ver. 2.6.1. Gene expression in transcripts per million was quantified using RSEM ver. 1.3.1 and parameters recommended by the GTEx project. Using the molecular functional picture (MFT) [20], we integrated the transcriptome data to classify each tumor sample into unique TME subgroups. The MFT technique was employed to calculate the expression signatures related to immune and pro-tumor processes, including effector and natural killer (NK) cells, as well as matrix remodeling.
For preprocessing and annotation, the scRNA-seq reads were aligned to the GRCh38 human genome reference using CellRanger ver. 5.0. Data from all samples were combined in R ver. 4.1 using the Seurat package ver. 4.0. Doublets were filtered using Scrublet ver. 0.1. Cells with low-quality libraries (< 500 genes) and high mitochondrial read proportions (> 20%) were excluded. Data from each sample were scaled and standardized, followed by principal component analysis. Batch correction was performed using Harmony ver. 1.2.1. We employed a shared nearest-neighbor modularity optimization-based clustering technique to identify cell clusters using a uniform manifold approximation and a projection (UMAP) algorithm for dimensionality reduction. The “FindAllMarkers” function in Seurat was used to identify and annotate different cell type clusters based on the expression of representative lineage markers [21,22]. Hierarchical relationships among cell types were assessed by calculating pairwise Spearman correlations from average expression profiles (derived using the Seurat “AverageExpression” function), and Euclidean distances were subsequently computed for clustering analysis. Monocle3 ver. 1.2.8 was used to visualize UMAPs and lineage relationships among macrophages and T lymphoid cells.
5. Predicting ploidy of epithelial cells
Copy number karyotyping of aneuploid tumors using CopyKAT ver. 1.0.5 was performed to distinguish malignant cells from normal epithelial cells.
6. Discovering transcriptional programs of macrophages
To identify robust gene programs (modules), a consensus non-negative matrix factorization (cNMF) was run on macrophages from each sample. Each run utilized an unprocessed matrix to construct a k-MM graph (K=9) to identify the top-expressed genes in the clusters (S1 Table). We then distinguished between the TAM-specific macrophages and normal macrophage-specific modules. To identify correlated gene programs, we calculated the Pearson correlation coefficient from the average gene expression levels across all samples. Module scores for top genes from macrophages in tumor, adjacent, and distant normal tissues were estimated using the “AddModuleScore” function in Seurat. Functional enrichment analysis for each module was conducted according to the Gene Ontology terms using ClusterProfiler ver. 4.2.2.
7. Color multiplex staining
For the 7-color multiplex staining, formalin-fixed paraffin-embedded blocks were utilized. An Opal 7-color manual IHC Kit (NEL 811001KT, PerkinElmer) was used for sequential staining. This protocol relies on fluorescent tyramide signal amplification reagents that retain their fluorescence after multiple treatments with a steamer designed to remove the primary and secondary antibodies. The slides were sequentially stained with the following antibodies and fluorescent dyes: anti-panCK (pan-cytokeratin, AE1/E3, Dako)/Opal-690; anti–programmed death-ligand 1 (anti–PD-L1; 22C3, Dako)/Opal-520; anti-CD68 (KP1, Dako)/Opal-620; anti-PD1 (NAT105, Biolegend)/Opal-780; anti-CD163 (EPR19518, Abcam)/Opal-570. After completing the sequential reactions, slides were counterstained with DAPI (FP1490, PerkinElmer) and mounted using Prolong Antifade fluorescence mounting medium (P36965, Invitrogen). Single-marker staining was performed to create a spectral library, which was analyzed to determine the exposure times for detecting specific signals. All the stained slides were stored in a light-proof environment at 4°C prior to imaging.
8. Imaging and tissue segmentation
Tissue segmentation was performed using inForm software, which was trained to identify tumor areas, stroma, and blank regions. Once an accurate algorithm for tissue segmentation was established, batch correction was conducted for all images. Cell segmentation was achieved by defining the nuclei to delineate the cell contours. A training-by-example interface was then used to identify cells on a cell-by-cell basis. Following cellular segmentation, cell phenotyping was conducted, in which positively stained target cells were identified using thresholds based on staining intensity, allowing the quantification of the number of positively stained target cells.
9. Nearest neighbor analysis
Nearest neighbor (NN) analysis was employed to quantify cell-cell interactions by calculating the distance between each cell of one phenotype and its nearest neighbor of another phenotype using the Phenoptr R package ver. 0.3.2. The NN distances and number of cells within a defined radius around each cell type were also calculated using Phenoptr R package ver. 0.3.2. The results were visualized by plotting the nearest neighbors, with cytokeratin-positive tumor cells represented in red and CD8+ T cells in blue, including the lines drawn from each tumor cell to its nearest CD8+ T cell.
10. Statistical analysis
Survival was analyzed using Kaplan-Meier plots, Cox proportional hazard models, and the log-rank test to determine p-values. All statistical analyses were conducted using R ver. 4.1.2 (R Foundation for Statistical Computing, http://www.R-project.org). Two different variables were compared using the Wilcoxon signed-rank test. All p-values were two-sided, and results were determined to be significant at p < 0.05. The relationship between variables was quantified using Pearson’s correlation, and the coefficient of determination (R) was calculated to assess the strength and direction of correlation.
1. Study population and genomic characteristics
Fifteen patients with mCRC who underwent lung metastasectomy with curative intent were enrolled in this study between October 2019 and December 2021 (S2A Fig.). WES was performed on 14 tumor tissues from 13 patients, whereas WTS was conducted on 18 matched tumors and adjacent normal tissues from all 15 patients. Two patients (P01 and P08) had multiple lung metastases resected (three and two, respectively). Multiplex IHC was performed on matched tumor and adjacent normal tissues from six patients, and scRNA-seq was conducted on the matched tumor and adjacent and distant normal tissues from five patients. The median age of the patients was 63 years (range, 49 to 77 years), and eight patients (53.3%) were male. All patients had MSS-CRC as the primary tumor. Thirteen patients (86.7%) had left-sided primary CRC, whereas two patients (13.3%) had right-sided primary CRC. Ten patients (67.7%) underwent wedge resection, and five patients (33.3%) underwent lobectomy. Among the 15 patients, two presented with lung metastases at the time of CRC diagnosis, eight developed lung metastases as the first site of recurrence, and five developed them as a second or subsequent recurrence. Fourteen patients had received systemic chemotherapy before lung metastasectomy, but all had a chemotherapy-free interval of at least 2.7 months (median, 10.7 months; range, 2.7 to 56.5 months). Eight patients (53.3%) experienced tumor recurrence on the follow-up date in July 2023: four in the resected lung, two in the bilateral lungs, one in the contralateral lung, and one in the ipsilateral pleura (Table 1, S3 Table).
In the WES analysis (n=13), mutations were most frequently found in TP53 (n=12, 92.3%), KRAS (n=10, 76.9%), and APC (n=10, 76.9%) (S2B Fig.). When applying the CRC-derived transcriptomic classification system, we found that the canonical (consensus molecular subtype 2 [CMS2]; n=9, 50%) and mesenchymal (CMS4; n=7, 38.9%) subtypes were the most prevalent, whereas the immune-activated (CMS1; n=1, 5.6%) and metabolic (CMS3; n=1, 5.6%) subtypes were rarely identified. Additionally, transcriptomic TME subtyping revealed enrichment of the immune-depleted (n=14, 77.8%) and fibrotic (n=4, 22.2%) subtypes, with no samples classified as immune-enriched.
2. Reduced function of effector T and NK cells in CRC lung metastases
To investigate the distinct TME features in CRC lung metastases, we compared immune gene expression between tumors and adjacent normal tissues (Fig. 1A). Tumor tissues exhibited significant downregulation of antitumor effector (p < 0.001), T (p < 0.01), and NK (p < 0.001) cell signatures (Fig. 1B and C). In contrast, tumor tissues showed significant upregulation of both matrix remodeling and proliferation signatures (p < 0.001) (S2C Fig.).
From the Multiplex-IHC data, we analyzed the number and spatial distribution of immune cells in CRC lung metastases (Fig. 2A and B). Cell phenotypes were determined according to stained markers; cell densities were measured by counting cells within a specific area (radius of either 15 or 30 μm), and cell proximities were assessed using the NN analysis (Fig. 2C). In tumor tissues, CD8+ T cells (p=0.015) and macrophages (CD68+, p < 0.01) were found at a significantly greater distance from epithelial cells than from adjacent normal tissues (Fig. 2D). Additionally, tumor tissue exhibited smaller densities of both T cells (p=0.036 and p=0.064 for a radius of 15 μm and 30 μm, respectively) and macrophages (p < 0.01 and p=0.015 for a radius of 15 μm and 30 μm, respectively) than adjacent normal tissue (Fig. 2E and F). In summary, effector immune cells (T and NK cells) were located at a distance from epithelial cells, with a limited presence in CRC lung metastases.
3. Increased regulatory T cells and decreased NK and T cell proportions in lung metastases
To identify the cell types associated with immunosuppression, we performed scRNA-seq to determine the cellular composition of the TME. A total of 50,238 cells were clustered in the UMAP plot, and global cell types were annotated based on marker gene expression (Fig. 3A, S4A and S4B Fig.). Hierarchical clustering analysis provided a global, lineage-level overview of transcriptional relationships among major cell populations across tumor, adjacent normal, and distant normal tissues (S4C Fig.).
This method allowed visualization of the hierarchical similarity structure among cell lineages, highlighting context-dependent transcriptional remodeling that may not be evident in UMAP projections. Epithelial cells in tumor tissues were completely separated from those in normal tissues, reflecting distinct lineage origins and gene-expression programs, whereas stromal cells were partially intermingled. Myeloid and lymphoid compartments were also partially segregated by tissue type, suggesting context-specific transcriptional adaptation even within shared immune lineages. Epithelial cells were more frequently found in tumor tissues than in adjacent or distant normal tissues, whereas stromal cells comprised a greater proportion of normal tissues than tumor tissues (Fig. 3B, S4B Fig.). The epithelial cell subtypes were further annotated using classical marker genes previously defined in the literature (S5A Fig.). Predicted aneuploid cells accounted for 35% of the total epithelial cells and 69% of the tumor epithelial cells, whereas none of the epithelial cells from adjacent or distant normal tissues were classified as aneuploid (S5B Fig.). Accordingly, malignant and proliferating epithelial cells constituted 20% and 5%, respectively, of the total cells from tumor tissues, with less than 0.1% found in normal tissues (S5C Fig.). Furthermore, CRC-enriched marker genes (CEACAM5, GPX2, PHGR1, and LGALS4) and colonic epithelial lineage markers (LGR5, OLFM4, SLC26A3, SPINK1, REG4, and MUC2) were frequently expressed in malignant and proliferating epithelial cells but not in other epithelial cell subtypes (S5D and S5E Fig.).
In tumor tissues, the proportion of B cells was higher and the proportion of plasma cells was significantly higher than those in the adjacent (p=0.015) and distant normal (p < 0.01) tissues (Fig. 3B and E). The proportion of NK cells was significantly lower in tumor tissues than in adjacent (p=0.026) and distant normal (p=0.032) tissues (Fig. 3B and E, S6A-S6D Fig.). While the overall proportion of T lymphoid cells was similar between tumor and normal tissues, regulatory T and type 1 T helper (Th1) CD4+ T cells were significantly more abundant in tumor tissues than in the adjacent (p < 0.01 regulatory T cell [Treg]; p=0.016, Th1) and distant normal tissues (p=0.016, Treg; p=0.016, Th1).
UMAP visualization of T and NK cell subpopulations (Fig. 3C), corresponding to S6A and S6C-F Fig., illustrates the lineage distribution across tumor, adjacent normal, and distant normal regions, serving as the basis for the functional comparisons in Fig. 3D and E. Conversely, effector CD8+ T and NK cells were significantly depleted in tumor tissues compared to levels in adjacent normal (p=0.060 for effector CD8+ and p=0.026 for NK cells) and distant normal (p=0.139 for effector CD8+ and p=0.032 for NK cells) tissues (Fig. 3D and E, S6A-S6F Fig.). Additionally, the proportions of effector and memory CD8+ T cells and NK cells were higher in the adjacent normal tissues than in the distant normal tissues.
To clarify the spatial patterns of CD8+ T-cell exhaustion, we further examined the T/NK-cell compartment (Fig. 3D, S6B Fig.). When all lymphoid populations were considered together, the relative proportion of exhausted CD8+ T cells appeared lower in tumors due to the increased abundance of B-lineage cells, particularly memory B and plasma cells, in tumor tissues. This compositional effect explains the apparent discrepancy observed in Fig. 3E. The expression patterns of exhaustion-associated genes (PDCD1, LAG3, and TIGIT) across tissue locations are presented in S6A and S6C Fig., which can be cross-referenced with clustering and marker-based annotation (S6F Fig.) shown in Fig. 3.
4. Enrichment of SPP1-expressing TAMs in lung metastases
TAMs were significantly enriched in tumor tissues compared to levels in adjacent normal tissues (p=0.016), while alveolar macrophages were significantly depleted in tumor tissues compared to those in adjacent (p=0.032) and distant normal (p < 0.01) tissues (Fig. 4A and B). TAMs consistently comprised approximately 40% of the total macrophages in tumor tissue, but only approximately 10% in normal tissues (S7A-S7C Fig.).
To identify the gene modules characterizing TAMs, we applied cNMF (Fig. 4C, S1 Table). Compared to TAMs in adjacent and distant normal tissues, TAMs in tumor tissues showed significant downregulation of gene module 1, which is associated with the immune response (S7D and S7E Fig.). In contrast, gene modules 5 and 7 were significantly upregulated in tumor-derived TAMs compared to those from normal tissues (p < 0.01 for all); gene-ontology analysis indicated that module 5 (e.g., TREM2, TMEM176B, and LIPA) was primarily linked to immune-regulatory processes, whereas module 7, containing SPP1, was enriched for Wnt signaling, inflammatory response, and cell migration, suggesting distinct transcriptional programs within TAMs.
Among module-defining genes, SPP1, TREM2, TMEM176B, and LIPA were selected because they exhibited the highest expression in tumor-derived TAMs and showed significant stepwise decreases with increasing distance from the tumor site (tumor > adjacent > distant; adjusted p < 0.01, |log2FC| > 0.5). Accordingly, SPP1 expression decreased progressively from tumor to adjacent and distant normal tissues (Fig. 4D and E), emphasizing its tumor-proximal enrichment and potential relevance to the immunosuppressive TME.
5. Correlation of SPP1-expressing TAMs with suppressed effector T cells
SPP1 expression in macrophages was significantly negatively correlated with the expression of marker genes for cytotoxic CD8+ T cells (R=–0.75, p < 0.01 with PRF1; R=–0.61, p=0.020 with GNLY) (Fig. 5A). Additionally, SPP1 expression in macrophages was positively correlated with exhausted CD8+ T cell markers (R=0.71, p < 0.01, for PDCD1; R=0.81, p < 0.01, for TIGIT; R=0.55, p=0.04, for LAG3) (Fig. 5B). Furthermore, it was positively correlated with regulatory CD4+ T cell markers (R=0.76, p < 0.01, for FOXP3; R=0.71, p < 0.01, for CTLA4) (Fig. 5C). Together, these findings suggest that SPP1+ macrophages are associated with attenuated cytotoxic activity and enhanced T-cell exhaustion within the tumor microenvironment.
To further evaluate whether this association was unique to SPP1, we performed the same correlation analysis for other macrophage-related genes (TREM2, TMEM176B, and LIPA) in S7F-H Fig. Gene Ontology analysis indicated that all three genes were associated with immune-regulatory and T cell–modulating functions. Consistent with this, all three genes showed negative correlations with cytotoxic markers (PRF1, GZMB), although only LIPA reached statistical significance. In the CD8+ compartment, all genes exhibited a trend toward positive correlation with exhaustion markers, but only PDCD1 and TIGIT achieved statistical significance across multiple genes. In the CD4+ compartment, LIPA was the only gene showing a significant positive correlation with exhaustion-related markers.
Overall, among the macrophage-associated genes examined, SPP1 displayed the most robust and consistent pattern— strongly negative with cytotoxicity and positive with exhaustion or Treg markers. These results highlight SPP1 as the macrophage-derived signal most tightly linked to T-cell dysfunction and tumor-proximal immunoregulation, supported by its stepwise spatial gradient (tumor > adjacent > distant) and correlation with T-cell gene signatures (Fig. 5). We were particularly interested in this feature because the tumor-adjacent normal tissue may represent a recurrenceprone niche after surgical resection, where residual immune remodeling could predispose to local relapse. In summary, SPP1-expressing macrophages define a distinct immunosuppressive axis that links macrophage activation to T-cell exhaustion within CRC lung metastases.
6. Association of upregulated SPP1 expression with recurrence in CRC lung metastases
We investigated the genes associated with recurrence by analyzing the tumor and adjacent normal tissues. Differentially upregulated genes in recurrent tumor tissues included SPP1, along with genes involved in matrix remodeling and angiogenesis (MMP9, PLVAP, FLT1, VWF, and PDGFA) (Fig. 6A). SPP1 expression was also upregulated in recurrent adjacent normal tissues along with MMP9 and CTLA4 expression (Fig. 6B), suggesting immunosuppressive roles. To further evaluate the prognostic relevance of SPP1 within our cohort (15 patients, 16 spatially distinct tissue samples), we performed a recurrence-free survival (RFS) analysis using bulk RNA-seq–derived transcript levels. Using the mstat R package, we determined an optimal cutpoint for SPP1 expression, quantified as transcripts per million, at 7.28 based on the log-rank test and classified samples into SPP1-high and SPP1-low groups. Patients with high SPP1 expression showed significantly shorter RFS, supporting the clinical relevance of SPP1 as a recurrence-associated macrophage marker (p=0.007) (Fig. 6C).
Moreover, SPP1 expression was significantly linked to tumor recurrence in both tumor and adjacent and distant normal tissues (p < 0.01 all) (Fig. 6D). Additionally, the proportion of exhausted CD8+ T cells among the total T and NK cells was associated with tumor recurrence, which was significant only in adjacent normal tissues (p=0.026) (S8A Fig.). Interestingly, the proportion of CD8+PD-1+ cells in the adjacent normal tissues, as measured by multiplex IHC, was also significantly associated with recurrence (Fig. 6E), highlighting the role of macrophages and T cells in tumor immune surveillance. To complement the proportion-based analyses, we assessed exhaustion-related transcriptional programs in T cells by calculating exhaustion scores using a curated gene set including PDCD1, ENTPD1, TNFRSF9, CXCL13, and TOX (S9 Table). As shown in S8B Fig., recurrent cases showed higher exhaustion scores in adjacent and distant normal tissues, consistent with the proportion-based results. Tumor regions exhibited uniformly elevated exhaustion scores, indicating that CD8+ T-cell dysfunction represents a core feature of the TME and extends systemically beyond the tumor site.
Collectively, our data suggest a model to explain recurrence after lung metastasectomy (Fig. 6F). The resected lung may be tumor-free, but already infiltrated with TAMs and Tregs, accompanied by the depletion and exhaustion of immune effector cells. An immunosuppressive environment is also associated with an increased risk of recurrence. These findings may inform the design of clinical trials targeting both TAMs and T cells for patients with CRC who have undergone lung metastasectomy.
This study provides a comprehensive analysis of the TME in CRC lung metastases, highlighting significant alterations in immune cell populations and their associations with disease recurrence. By leveraging WES, WTS, scRNA-seq, and multiplex IHC, we elucidated the key features of the lung TME that may influence the prognosis of patients with CRC. Although the sample size was limited (15 patients), the integration of multi-omic and spatial analyses yielded consistent immune signatures. scRNA-seq, performed in five patients including distant normal tissues, provided key insights but limited power for detailed cell-cell interaction analyses, warranting validation in larger cohorts.
A key finding was the enrichment of SPP1-expressing TAMs in tumor tissues. To the best of our knowledge, this is the first report of SPP1-expressing TAM enrichment in CRC lung metastases. SPP1 (osteopontin) expression in macrophages is strongly correlated with markers of T cell exhaustion and Tregs and negatively correlated with cytotoxic T cell markers. Among the macrophage subtypes, only SPP1-expressing TAMs were more abundant in lung metastases than in normal tissues, and SPP1 expression was associated with recurrence in tumor tissues, as well as in adjacent and distant normal tissues. The proportion of exhausted T cells in normal tissues adjacent to lung metastases and Treg developmental pathway score in normal tissues adjacent to mCRC were also associated with recurrence. This is consistent with the results of previous studies showing that SPP1-expressing TAMs play an immunosuppressive role in various cancers [23]. Previous research has also shown enrichment of SPP1-expressing TAMs in liver metastases compared to in primary CRCs [24,25].
Our findings corroborate those of recent studies on CRC metastases that identified immunosuppressive changes in the TME characterized by SPP1-expressing macrophages and Treg cells [26]. SPP1 expression in macrophages was negatively correlated with the expression of cytotoxic genes in CD8+ T cells but positively correlated with the expression of CD8+ T cell exhaustion and CD4+ T cell immunoregulatory genes. We observed depletion of NK and effector CD8+ T cells in the tumor tissue, accompanied by an increase in Tregs. These changes indicate an immunosuppressive environment that may facilitate tumor growth and metastasis [23]. In a CT26 murine colon cancer model, neutralizing SPP1 with a monoclonal antibody increased the lytic activity of cytotoxic T cells in vitro and augmented the efficacy of anti-PD-1 treatment in vivo [27].
scRNA-seq analysis revealed significant differences in immune cell populations between tumor and normal tissues. SPP1 is associated with tumor recurrence not only in tumor tissues but also in adjacent normal tissues. The enrichment of these cell types in both tumor and adjacent normal tissues may explain why some patients experience recurrence, even after complete surgical resection. The observation that these changes extend to the adjacent normal tissue suggests that the immunosuppressive effects of the tumor extend beyond its visible borders. This highlights the importance of considering a broader TME when developing treatment strategies for mCRC.
Our study also demonstrated that TME features may have prognostic significance. The proportion of exhausted CD8+ T cells and expression of PD-1 in these cells in adjacent normal tissues are associated with tumor recurrence. This is consistent with previous studies showing that T cell exhaustion is a key mechanism of immune evasion in cancer. Furthermore, the association between SPP1 expression and poor overall survival in an independent cohort of patients with primary CRC underscores its potential as a prognostic biomarker [22].
The presence of an immunosuppressive TME, characterized by SPP1-expressing TAMs and exhausted T cells, may contribute to the limited efficacy of current immunotherapies for MSS CRC [23]. Our results suggest that combination therapies targeting both TAMs and T cells may be necessary to overcome this immunosuppressive environment and improve outcomes in patients with mCRC. Although SPP1+ TAMs were strongly associated with T-cell exhaustion and immunosuppression, this study did not include functional blockade or depletion experiments to confirm causality. This limitation has been acknowledged, and future mechanistic studies are needed to validate SPP1-directed therapeutic strategies.
While several SPP1 modulation strategies have been proposed—including small-molecule inhibitors, siRNA/miRNA approaches, OPN-neutralizing antibodies, and disruption of OPN–integrin or CD44 interactions—direct pharmacologic inhibition of SPP1 remains at the preclinical stages [28]. Recent work in prostate cancer demonstrated that SPP1+ TAMs suppress CD8+ T-cell function via the adenosine signaling pathway; in that context, dual blockade of the adenosine A2A receptor and PD-L1 restored antitumor [29].
The clinical benefit of PD-1 inhibitors in MSS CRC remains poor, underscoring the need for new approaches to overcome resistance. Notably, the lung and liver, two predominant metastatic sites of CRC, harbor distinct immune microenvironments, suggesting that organ-specific factors may influence immunotherapeutic responsiveness.
Taken together, our observations raise the possibility that combined modulation of the SPP1 pathway and immune checkpoint blockade could be considered in future studies, particularly in MSS CRC patients with pulmonary metastases or in the peri-metastasectomy context, where residual immunosuppressive niches are likely to remain despite complete resection. Moreover, the observation that the immunosuppressive TME extends to the adjacent normal tissues also has implications for surgical management. This suggests that wider resection margins are necessary to remove all potentially affected tissues. Additionally, neoadjuvant immunotherapy or targeted therapies aimed at reversing the immunosuppressive TME could improve surgical outcomes [30].
In conclusion, our study provides a comprehensive characterization of the TME in CRC lung metastases and reveals an immunosuppressive environment mediated by multiple cell types. These findings provide new insights into the mechanisms of metastasis and CRC recurrence and suggest potential therapeutic targets. Future studies should focus on the development and testing of combination therapies that can effectively target both tumor cells and the immunosuppressive TME, including effector T cells and TAMs, to improve outcomes in patients with mCRC.
Supplementary materials are available at Cancer Research and Treatment website (https://www.e-crt.org).
crt-2025-691_S1_Table.xlsx
crt-2025-691_S2_Fig.pdf
crt-2025-691_S3_Table.xlsx
crt-2025-691_S4_Fig.pdf
crt-2025-691_S5_Fig.pdf
crt-2025-691_S6_Fig.pdf
crt-2025-691_S7_Fig.pdf
crt-2025-691_S8_Fig.pdf
crt-2025-691_S9_Table.pdf

Ethical Statement

This study was approved by the Institutional Review Board of Samsung Medical Center (IRB File No. 2021-10-054). Written informed consent was obtained from all patients.

Author Contributions

Conceived and designed the analysis: Choi YS, Lee J.

Collected the data: Jeon YJ, Hong TH.

Contributed data or analysis tools: Shin JK, Lim SH, Park Y, Cho YB, Kim ST.

Performed the analysis: Kwon M, Shin MK, An M.

Wrote the paper: Kwon M, Shin MK, An M.

Conflicts of Interest

Conflict of interest relevant to this article was not reported.

Funding

This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (grant number: RS-2023-00222838 to JL) and the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health & Welfare, Republic of Korea (grant number HI23C1589 to MK).

Fig. 1.
Transcriptional and immune profiles of colorectal cancer lung metastases. (A) Workflow illustrating the multi-region sampling strategy for metastatic colon cancer and spatially defined normal lung for multi-omics analysis. (B) Heatmap displaying relative signature scores, including tumor microenvironment composition. (C) Box plot showing scores of anti-tumor-related immune signatures between tumor (n=18) and adjacent normal tissues (n=16) (Wilcoxon matched-pairs signed-rank test). CAF, cancer-associated fibroblast; DC, dendritic cell; EMT, epithelial-mesenchymal transition; MDSC, myeloid-derived suppressor cell; NK, natural killer; Th1, T helper 1; Th2, T helper 2; Treg, regulatory T cell.
crt-2025-691f1.jpg
Fig. 2.
Immune exclusion of metastatic tumors. (A) Multiplex immunofluorescence (mIF) image of the metastatic lung from a representative patient (S06). (B) mIF image of adjacent normal lung from the same patient (S06). PD-L1, programmed death-ligand 1; PD-1, programmed death-1. (C) Schematic illustration of the spatial analysis of multiplex immunofluorescence images. (D) Comparison of the proximity of classic T cells and classic macrophages from cytokeratin-positive (CK+) cells across metastatic tumor (n=6) and adjacent normal samples (n=6). (E) Comparison of CD8+ T cell density within radii of 15 μm and 30 μm across metastatic tumor and adjacent normal samples. (F) Comparison of CD68+ macrophage density within radii of 15 μm and 30 μm across metastatic tumor and adjacent normal samples. All p-values were determined using the two-sided Wilcoxon signed-rank test (D-F).
crt-2025-691f2.jpg
Fig. 3.
Spatially heterogeneous microenvironment of colorectal cancer lung metastases. (A) Uniform manifold approximation and projection (UMAP) visualization of tumor, adjacent normal, and distant normal samples, colored by annotated global cell types. (B) Stacked bar plot illustrating different cellular compositions across three spatial locations (tumor, adjacent normal, and distant normal) obtained from five patients. The distant-normal sample from P08 was excluded due to sequencing failure. (C) UMAP visualization of T and natural killer (NK) cells, colored by subclassification. (D) Stacked bar plot illustrating the distribution of T and NK cell subpopulations across tumor, adjacent normal, and distant normal regions from five patients. Treg, regulatory T cell. (E) Comparison of lymphoid subpopulation proportions among various locations (two-sided Wilcoxon signed-rank test).
crt-2025-691f3.jpg
Fig. 4.
Tumor-associated biomarkers in macrophages. (A) Uniform manifold approximation and projection (UMAP) visualization of macrophages, colored by subclassification. (B) Comparison of myeloid subpopulation proportions among different locations (tumor, adjacent normal, and distant normal) obtained from five patients. (two-sided Wilcoxon signed-rank test). (C) Heatmap showing pairwise correlation of gene modules across all samples using the consensus non-negative matrix factorization algorithm. (D) Four tumor-associated macrophage (TAM) markers showing decreasing expression from tumor to distant normal tissues (one-way ANOVA). (E) Feature plot depicting four TAM markers on UMAP. AN, adjacent normal; DN, distant normal.
crt-2025-691f4.jpg
Fig. 5.
Correlation of SPP1 expression in macrophages with immunosuppressive T cell phenotypes. (A) Scatter plot showing Spearman correlation coefficients between SPP1 in macrophages and PRF1 and GNLY in CD8+ T cells across tumor, adjacent, and distant normal tissues. (B) Scatter plot showing Spearman correlation coefficients between SPP1 in macrophages and PDCD1, TIGIT, and LAG3 in CD8+ T cells across the same tissues. (C) Scatter plot showing Spearman correlation coefficients between SPP1 in macrophages and FOXP3 and CTLA4 in CD4+ T cells across the same tissues.
crt-2025-691f5.jpg
Fig. 6.
SPP1-expressing macrophages as significant contributors to recurrent pulmonary metastases. (A) Volcano plot showing differential gene expression between recurrent and non-recurrent tumor tissues. (B) Volcano plot showing differential gene expression between recurrent and non-recurrent adjacent normal tissues. (C) Kaplan-Meier curves of recurrence-free survival (RFS) for high- and low-SPP1 expression groups in this metastatic cohort (n=15, 16 samples). mRFS, median RFS; TPM, transcripts per million. For patient P08, two spatially distinct adjacent-normal samples were individually analyzed (Log-rank test). (D) Box plot comparing SPP1 expression levels from single-cell RNA sequencing between recurrent and non-recurrent samples across different tissue locations (two-sided Wilcoxon signed-rank test). (E) Representative multiplex immunofluorescence (mIF) images of adjacent normal tissues from recurrent (S08) and non-recurrent (S04) cases (left). Comparison of CD8+PD-1+ proportions calculated from mIF images (two-sided Wilcoxon signed-rank test) (right). (F) Proposed model of the tumor microenvironment in recurrent versus non-recurrent cases during lobectomy from patients with colorectal cancer lung metastases. CK, cytokeratin; NK, natural killer; PD-L1, programmed death-ligand 1; PD-1, programmed death-1; Treg, regulatory T cell.
crt-2025-691f6.jpg
Table 1.
Patient characteristics
Clinical characteristic Value (n=15)
Age (yr) 63 (49-77)
Sex
 Male 8 (53.3)
 Female 7 (46.7)
Colon side
 Left 13 (86.7)
 Right 2 (13.3)
Location
 Ascending colon 2 (13.3)
 Sigmoid 4 (26.7)
 Rectosigmoid 1 (6.7)
 Rectum 7 (46.7)
 Transverse colon and rectum 1 (6.7)
Surgical procedure
 Lobectomy 5 (33.3)
 Wedge resection 10 (66.7)
Recurrence
 Recurrence 8 (53.3)
 Non-recurrence 7 (46.7)

Values are presented as median (range) or number (%).

  • 1. Choti MA, Sitzmann JV, Tiburi MF, Sumetchotimetha W, Rangsin R, Schulick RD, et al. Trends in long-term survival following liver resection for hepatic colorectal metastases. Ann Surg. 2002;235:759–66. ArticlePubMedPMC
  • 2. Nordlinger B, Sorbye H, Glimelius B, Poston GJ, Schlag PM, Rougier P, et al. Perioperative chemotherapy with FOLFOX4 and surgery versus surgery alone for resectable liver metastases from colorectal cancer (EORTC Intergroup trial 40983): a randomised controlled trial. Lancet. 2008;371:1007–16. ArticlePubMedPMC
  • 3. Cervantes A, Adam R, Rosello S, Arnold D, Normanno N, Taieb J, et al. Metastatic colorectal cancer: ESMO clinical practice guideline for diagnosis, treatment and follow-up. Ann Oncol. 2023;34:10–32. ArticlePubMed
  • 4. Carpizo DR, D’Angelica M. Liver resection for metastatic colorectal cancer in the presence of extrahepatic disease. Ann Surg Oncol. 2009;16:2411–21. ArticlePubMed
  • 5. Antoniotti C, Rossini D, Pietrantonio F, Catteau A, Salvatore L, Lonardi S, et al. Upfront FOLFOXIRI plus bevacizumab with or without atezolizumab in the treatment of patients with metastatic colorectal cancer (AtezoTRIBE): a multicentre, open-label, randomised, controlled, phase 2 trial. Lancet Oncol. 2022;23:876–87. ArticlePubMed
  • 6. Chen EX, Jonker DJ, Loree JM, Kennecke HF, Berry SR, Couture F, et al. Effect of combined immune checkpoint inhibition vs best supportive care alone in patients with advanced colorectal cancer: the Canadian Cancer Trials Group CO.26 study. JAMA Oncol. 2020;6:831–8. ArticlePubMedPMC
  • 7. Tabernero J, Grothey A, Van Cutsem E, Yaeger R, Wasan H, Yoshino T, et al. Encorafenib plus cetuximab as a new standard of care for previously treated BRAF V600E-mutant metastatic colorectal cancer: updated survival results and subgroup analyses from the BEACON study. J Clin Oncol. 2021;39:273–84. ArticlePubMedPMC
  • 8. Raghav K, Siena S, Takashima A, Kato T, Van den Eynde M, Pietrantonio F, et al. Trastuzumab deruxtecan in patients with HER2-positive advanced colorectal cancer (DESTINYCRC02): primary results from a multicentre, randomised, phase 2 trial. Lancet Oncol. 2024;25:1147–62. ArticlePubMed
  • 9. Andre T, Shiu K, Kim TW, Jensen BV, Jensen LH, Punt C, et al. Pembrolizumab in microsatellite-instability-high advanced colorectal cancer. N Engl J Med. 2020;383:2207–18. PubMed
  • 10. Yan S, Wang W, Feng Z, Xue J, Liang W, Wu X, et al. Immune checkpoint inhibitors in colorectal cancer: limitation and challenges. Front Immunol. 2024;15:1403533.ArticlePubMedPMC
  • 11. Fukuoka S, Hara H, Takahashi N, Kojima T, Kawazoe A, Asayama M, et al. Regorafenib plus nivolumab in patients with advanced gastric or colorectal cancer: an open-label, dose-escalation, and dose-expansion phase Ib trial (REGONIVO, EPOC1603). J Clin Oncol. 2020;38:2053–61. ArticlePubMed
  • 12. Wang Q, Shao X, Zhang Y, Zhu M, Wang FXC, Mu J, et al. Role of tumor microenvironment in cancer progression and therapeutic strategy. Cancer Med. 2023;12:11149–65. ArticlePubMedPMCPDF
  • 13. Tumeh PC, Hellmann MD, Hamid O, Tsai KK, Loo KL, Gubens MA, et al. Liver metastasis and treatment outcome with anti-PD-1 monoclonal antibody in patients with melanoma and NSCLC. Cancer Immunol Res. 2017;5:417–24. ArticlePMCPDF
  • 14. Landi L, D’Inca F, Gelibter A, Chiari R, Grossi F, Delmonte A, et al. Bone metastases and immunotherapy in patients with advanced non-small-cell lung cancer. J Immunother Cancer. 2019;7:316.ArticlePubMedPMCPDF
  • 15. Pires da Silva I, Lo S, Quek C, Gonzalez M, Carlino MS, Long GV, et al. Site-specific response patterns, pseudoprogression, and acquired resistance in patients with melanoma treated with ipilimumab combined with anti-PD-1 therapy. Cancer. 2020;126:86–97. ArticlePubMedPDF
  • 16. Conway JW, Braden J, Wilmott JS, Scolyer RA, Long GV, Pires da Silva I. The effect of organ-specific tumor microenvironments on response patterns to immunotherapy. Front Immunol. 2022;13:1030147.ArticlePubMedPMC
  • 17. Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics. 2010;26:589–95. ArticlePubMedPMCPDF
  • 18. Cibulskis K, Lawrence MS, Carter SL, Sivachenko A, Jaffe D, Sougnez C, et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol. 2013;31:213–9. ArticlePubMedPMCPDF
  • 19. Kim S, Scheffler K, Halpern AL, Bekritsky MA, Noh E, Kallberg M, et al. Strelka2: fast and accurate calling of germline and somatic variants. Nat Methods. 2018;15:591–4. ArticlePubMedPDF
  • 20. Bagaev A, Kotlov N, Nomie K, Svekolkin V, Gafurov A, Isaeva O, et al. Conserved pan-cancer microenvironment subtypes predict response to immunotherapy. Cancer Cell. 2021;39:845–65. Article
  • 21. Lee H, Hong Y, Etlioglu HE, Cho YB, Pomella V, Van den Bosch B, et al. Lineage-dependent gene expression programs influence the immune landscape of colorectal cancer. Nat Genet. 2020;52:594–603. ArticlePDF
  • 22. Zhang L, Li Z, Skrzypczynska KM, Fang Q, Zhang W, O’Brien SA, et al. Single-cell analyses inform mechanisms of myeloid-targeted therapies in colon cancer. Cell. 2020;181:442–59. Article
  • 23. Zhang Y, Rajput A, Jin N, Wang J. Mechanisms of immunosuppression in colorectal cancer. Cancers (Basel). 2020;12:3850.ArticlePubMed
  • 24. Sathe A, Mason K, Grimes SM, Zhou Z, Lau BT, Bai X, et al. Colorectal cancer metastases in the liver establish immunosuppressive spatial networking between tumor-associated SPP1+ macrophages and fibroblasts. Clin Cancer Res. 2023;29:244–60. ArticlePubMedPMCPDF
  • 25. Qi J, Sun H, Zhang Y, Wang Z, Xun Z, Li Z, et al. Single-cell and spatial analysis reveal interaction of FAP(+) fibroblasts and SPP1(+) macrophages in colorectal cancer. Nat Commun. 2022;13:1742.ArticlePubMedPDF
  • 26. Zafari N, Khosravi F, Rezaee Z, Esfandyari S, Bahiraei M, Bahramy A, et al. The role of the tumor microenvironment in colorectal cancer and the potential therapeutic approaches. J Clin Lab Anal. 2022;36:e24585ArticlePubMedPMCPDF
  • 27. Klement JD, Poschel DB, Lu C, Merting AD, Yang D, Redd PS, et al. Osteopontin blockade immunotherapy increases cytotoxic T lymphocyte lytic activity and suppresses colon tumor progression. Cancers (Basel). 2021;13:1006.ArticlePubMedPMC
  • 28. Liu W, Kuang T, Liu L, Deng W. The role of innate immune cells in the colorectal cancer tumor microenvironment and advances in anti-tumor therapy research. Front Immunol. 2024;15:1407449.ArticlePMC
  • 29. Lyu A, Fan Z, Clark M, Lea A, Luong D, Setayesh A, et al. Evolution of myeloid-mediated immunotherapy resistance in prostate cancer. Nature. 2025;637:1207–17. ArticlePubMedPDF
  • 30. Yan Z, Hu X, Tang B, Deng F. Role of osteopontin in cancer development and treatment. Heliyon. 2023;9:e21055ArticlePubMedPMC

Figure & Data

REFERENCES

    Citations

    Citations to this article as recorded by  
    • Obesity promotes aggressiveness of endometrial cancer via metabolic reprogramming and intercellular crosstalk in the tumor microenvironment
      Yu Chen, Yijin Fang, Guozhi Zhao, Yuanqun Zhou, Dong-Hua Yang, Jian Han, Ying Zheng
      Cancer Letters.2026; 656: 218649.     CrossRef

    • PubReader PubReader
    • ePub LinkePub Link
    • Cite
      CITE
      export Copy Download
      Close
      Download Citation
      Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.
      Format:
      • RIS — For EndNote, ProCite, RefWorks, and most other reference management software
      • BibTeX — For JabRef, BibDesk, and other BibTeX-specific software
      Include:
      • Citation for the content below
      Immunosuppressive Tumor Microenvironment in Colorectal Cancer Lung Metastases: Implications for Recurrence after Metastasectomy
      Close
    • XML DownloadXML Download
    • 0
    • 1
    • 2
    • 3
    • 4
    • 5
    Immunosuppressive Tumor Microenvironment in Colorectal Cancer Lung Metastases: Implications for Recurrence after Metastasectomy
    Image Image Image Image Image Image
    Fig. 1. Transcriptional and immune profiles of colorectal cancer lung metastases. (A) Workflow illustrating the multi-region sampling strategy for metastatic colon cancer and spatially defined normal lung for multi-omics analysis. (B) Heatmap displaying relative signature scores, including tumor microenvironment composition. (C) Box plot showing scores of anti-tumor-related immune signatures between tumor (n=18) and adjacent normal tissues (n=16) (Wilcoxon matched-pairs signed-rank test). CAF, cancer-associated fibroblast; DC, dendritic cell; EMT, epithelial-mesenchymal transition; MDSC, myeloid-derived suppressor cell; NK, natural killer; Th1, T helper 1; Th2, T helper 2; Treg, regulatory T cell.
    Fig. 2. Immune exclusion of metastatic tumors. (A) Multiplex immunofluorescence (mIF) image of the metastatic lung from a representative patient (S06). (B) mIF image of adjacent normal lung from the same patient (S06). PD-L1, programmed death-ligand 1; PD-1, programmed death-1. (C) Schematic illustration of the spatial analysis of multiplex immunofluorescence images. (D) Comparison of the proximity of classic T cells and classic macrophages from cytokeratin-positive (CK+) cells across metastatic tumor (n=6) and adjacent normal samples (n=6). (E) Comparison of CD8+ T cell density within radii of 15 μm and 30 μm across metastatic tumor and adjacent normal samples. (F) Comparison of CD68+ macrophage density within radii of 15 μm and 30 μm across metastatic tumor and adjacent normal samples. All p-values were determined using the two-sided Wilcoxon signed-rank test (D-F).
    Fig. 3. Spatially heterogeneous microenvironment of colorectal cancer lung metastases. (A) Uniform manifold approximation and projection (UMAP) visualization of tumor, adjacent normal, and distant normal samples, colored by annotated global cell types. (B) Stacked bar plot illustrating different cellular compositions across three spatial locations (tumor, adjacent normal, and distant normal) obtained from five patients. The distant-normal sample from P08 was excluded due to sequencing failure. (C) UMAP visualization of T and natural killer (NK) cells, colored by subclassification. (D) Stacked bar plot illustrating the distribution of T and NK cell subpopulations across tumor, adjacent normal, and distant normal regions from five patients. Treg, regulatory T cell. (E) Comparison of lymphoid subpopulation proportions among various locations (two-sided Wilcoxon signed-rank test).
    Fig. 4. Tumor-associated biomarkers in macrophages. (A) Uniform manifold approximation and projection (UMAP) visualization of macrophages, colored by subclassification. (B) Comparison of myeloid subpopulation proportions among different locations (tumor, adjacent normal, and distant normal) obtained from five patients. (two-sided Wilcoxon signed-rank test). (C) Heatmap showing pairwise correlation of gene modules across all samples using the consensus non-negative matrix factorization algorithm. (D) Four tumor-associated macrophage (TAM) markers showing decreasing expression from tumor to distant normal tissues (one-way ANOVA). (E) Feature plot depicting four TAM markers on UMAP. AN, adjacent normal; DN, distant normal.
    Fig. 5. Correlation of SPP1 expression in macrophages with immunosuppressive T cell phenotypes. (A) Scatter plot showing Spearman correlation coefficients between SPP1 in macrophages and PRF1 and GNLY in CD8+ T cells across tumor, adjacent, and distant normal tissues. (B) Scatter plot showing Spearman correlation coefficients between SPP1 in macrophages and PDCD1, TIGIT, and LAG3 in CD8+ T cells across the same tissues. (C) Scatter plot showing Spearman correlation coefficients between SPP1 in macrophages and FOXP3 and CTLA4 in CD4+ T cells across the same tissues.
    Fig. 6. SPP1-expressing macrophages as significant contributors to recurrent pulmonary metastases. (A) Volcano plot showing differential gene expression between recurrent and non-recurrent tumor tissues. (B) Volcano plot showing differential gene expression between recurrent and non-recurrent adjacent normal tissues. (C) Kaplan-Meier curves of recurrence-free survival (RFS) for high- and low-SPP1 expression groups in this metastatic cohort (n=15, 16 samples). mRFS, median RFS; TPM, transcripts per million. For patient P08, two spatially distinct adjacent-normal samples were individually analyzed (Log-rank test). (D) Box plot comparing SPP1 expression levels from single-cell RNA sequencing between recurrent and non-recurrent samples across different tissue locations (two-sided Wilcoxon signed-rank test). (E) Representative multiplex immunofluorescence (mIF) images of adjacent normal tissues from recurrent (S08) and non-recurrent (S04) cases (left). Comparison of CD8+PD-1+ proportions calculated from mIF images (two-sided Wilcoxon signed-rank test) (right). (F) Proposed model of the tumor microenvironment in recurrent versus non-recurrent cases during lobectomy from patients with colorectal cancer lung metastases. CK, cytokeratin; NK, natural killer; PD-L1, programmed death-ligand 1; PD-1, programmed death-1; Treg, regulatory T cell.

    Fig. 1.

    Fig. 2.

    Fig. 3.

    Fig. 4.

    Fig. 5.

    Fig. 6.

    Immunosuppressive Tumor Microenvironment in Colorectal Cancer Lung Metastases: Implications for Recurrence after Metastasectomy
    Clinical characteristic Value (n=15)
    Age (yr) 63 (49-77)
    Sex
     Male 8 (53.3)
     Female 7 (46.7)
    Colon side
     Left 13 (86.7)
     Right 2 (13.3)
    Location
     Ascending colon 2 (13.3)
     Sigmoid 4 (26.7)
     Rectosigmoid 1 (6.7)
     Rectum 7 (46.7)
     Transverse colon and rectum 1 (6.7)
    Surgical procedure
     Lobectomy 5 (33.3)
     Wedge resection 10 (66.7)
    Recurrence
     Recurrence 8 (53.3)
     Non-recurrence 7 (46.7)
    Table 1. Patient characteristics

    Values are presented as median (range) or number (%).

    Table 1.

    Cancer Res Treat : Cancer Research and Treatment
    © Korean Cancer Association.
    Close layer
    TOP