Colorectal cancer (CRC) is the third leading cause of cancer death in both men and women. About 80-85% of all CRC cases are classified as microsatellite-stable (MSS), meaning they lack a particular type of DNA repair defect that makes tumors highly immunogenic (recognizable by the immune system).
MSS tumors are often described as immunologically "cold" - they generate few tumor-specific mutations that the immune system can recognize, and they actively exclude or inactivate the immune cells that do try to enter the tumor. This makes them largely resistant to immunotherapy drugs like checkpoint inhibitors that have transformed treatment for other cancers.
The approximately 15-20% of CRC tumors that are microsatellite-instable (MSI) behave very differently - they generate many more mutant proteins that act as targets for the immune system, making them highly responsive to immunotherapy. This stark contrast highlights why understanding the tumor immune microenvironment in MSS tumors is so critical.
Despite being considered immune cold, MSS CRC is actually a diverse group of tumors with varying degrees of immune infiltration. Finding ways to identify the subset of MSS tumors that might respond to immune-based approaches - or to design strategies to make cold tumors hot - represents a major research priority.
PD-L1 (programmed death-ligand 1) is a protein that can be displayed on the surface of both cancer cells and immune cells within a tumor. When PD-L1 binds to the PD-1 receptor on T cells, it puts the brakes on the immune response - a mechanism that normally prevents autoimmune damage but that tumors can exploit to evade destruction.
Counterintuitively, when PD-L1 is expressed on immune cells rather than tumor cells, it often reflects an active, ongoing immune battle within the tumor - a sign that immune cells have been called in and that the tumor is trying to shut them down. This immune-cell PD-L1 expression can be a marker of an immunologically active tumor environment.
In contrast, tumor cell PD-L1 expression may indicate a more immunosuppressive state or may simply reflect a tumor cell's intrinsic defense mechanisms. Distinguishing where PD-L1 comes from - tumor cells versus immune cells - is therefore critical for interpreting its prognostic meaning in any given cancer type.
Prior studies in CRC have given inconsistent results for PD-L1 as a prognostic marker, partly because they often mixed MSI and MSS patients together. Since MSI tumors inherently have higher PD-L1 expression, separating the two groups is essential for understanding PD-L1's true role in MSS tumors specifically.
This study analyzed tumor tissue from 254 patients with localized MSS colorectal cancer who underwent curative surgery at a Spanish university hospital. Patients were collected both retrospectively (from 2015-2017) and prospectively (from 2021-2023), providing a broad and diverse dataset.
The researchers used immunohistochemistry (IHC) - a technique that uses antibodies to stain specific proteins in tumor tissue sections - to measure the density of multiple immune cell types: CD3+ T cells, CD4+ helper T cells, CD8+ cytotoxic T cells, CD163+ macrophages, FoxP3+ regulatory T cells, CD20+ B cells, and PD-L1.
For deeper molecular analysis, the team used bulk RNA sequencing on 124 tumor samples to measure which genes were more or less active in PD-L1-positive versus PD-L1-negative tumors. An EMT score was also calculated for each tumor to determine whether it had an epithelial (less invasive) or mesenchymal (more invasive and migratory) gene expression profile.
The most advanced technique used was Visium HD spatial transcriptomics on four representative cases - a cutting-edge platform that measures gene activity in thousands of tiny spots across a tissue section while preserving the exact physical location of each measurement. This allowed researchers to map gene expression patterns to specific microenvironments within the tumor.
Tertiary lymphoid structures (TLSs) are organized clusters of immune cells that form within tumors, resembling miniature lymph nodes. They represent sites where the adaptive immune system is actively learning to fight the tumor. TLSs were found in 71-90% of MSS CRC cases in this study (depending on the detection method used).
Tumors with TLSs had significantly higher densities of CD3+ and CD4+ T cells at the tumor's invasive front - the boundary where cancer cells are actively pushing into surrounding normal tissue. Interestingly, TLS-positive tumors had lower densities of CD163+ macrophages at the invasive front, suggesting that an active adaptive immune response and abundant M2-type macrophages tend to be mutually exclusive.
Within TLS-positive tumors, those with mature TLSs (containing organized germinal centers marked by CD23+ follicular dendritic cells) had higher CD4+ and CD8+ T cell infiltration than tumors with only immature TLSs. This suggests that more organized lymphoid structures support more effective immune activity.
Unexpectedly, TLS content was highest in CMS4 tumors - the mesenchymal subtype known for poor prognosis. This suggests that even in tumors with features of immune evasion, some level of adaptive immune response is being mounted, which may create a therapeutic opportunity.
PD-L1 expression on immune cells was found in 70% of cases, predominantly within lymphoid aggregates at the tumor's invasive front. In stark contrast, PD-L1 expression on tumor cells themselves was rare (only 3% of cases) and showed no association with patient outcomes.
In survival analysis of 185 cases with follow-up data, immune cell PD-L1 expression was the only tumor microenvironment factor significantly associated with disease-free survival. Patients with PD-L1-positive immune cells had dramatically better outcomes: 5-year disease-free survival of 73% versus 52% in PD-L1-negative cases (hazard ratio of 0.18, meaning their risk of relapse was reduced by 82%).
When PD-L1 status was combined with clinical stage in a combined risk stratification model, the results were striking: PD-L1-negative patients at early stage I-II had a 15-fold higher risk of relapse than PD-L1-positive stage I-II patients. This means PD-L1 negativity at an early stage may actually confer a worse prognosis than PD-L1 positivity at a more advanced stage.
PD-L1 remained an independent predictor of disease-free survival even after accounting for lymph node involvement, perineural invasion, lymphovascular invasion, tumor depth, and tumor budding in multivariable analysis - confirming it provides unique prognostic information beyond standard clinical factors.
Bulk RNA sequencing revealed fundamentally different gene activity patterns in PD-L1-positive versus PD-L1-negative tumors. PD-L1-positive tumors showed overexpression of genes involved in immune responses, including lectin receptors and inflammatory molecules, reflecting an immunologically active state.
PD-L1-negative tumors showed enrichment of genes related to cell adhesion, cytoskeletal reorganization, and invasiveness - including genes like ADAM15, THBS1, VIM, and ACTG1 that are linked to epithelial-to-mesenchymal transition (EMT). EMT is a process where tumor cells acquire properties of more migratory, invasive cell types.
An EMT scoring analysis confirmed that 69% of PD-L1-negative tumors had a mesenchymal gene expression profile, while epithelial tumors were predominantly PD-L1-positive. This statistical link between PD-L1 negativity and mesenchymal identity is a key finding of the study.
PD-L1-negative tumors were also more frequently classified as the CMS4 molecular subtype (50% of PD-L1-negative cases), the consensus molecular subtype characterized by marked mesenchymal and stromal features and the worst overall prognosis among the four colorectal cancer subtypes.
Spatial transcriptomics allowed the researchers to map gene expression across specific regions of tumor tissue. When comparing macrophage-enriched stromal areas between PD-L1-positive and PD-L1-negative tumors, dramatically different molecular environments were revealed.
In PD-L1-positive tumors, macrophage-rich areas overexpressed genes associated with innate immunity and antigen presentation (LYZ, CD74, CST3, B2M, TAPBP) along with complement system components (C3, C1R, C1S) and tissue remodeling regulators (TIMP1, DCN). These genes collectively indicate macrophages that are engaged in immune detection and tumor surveillance.
In PD-L1-negative tumors, the same macrophage-rich areas looked completely different at the molecular level - dominated by collagen genes (COL1A1, COL3A1, COL5A1), cytoskeletal proteins (FN1, VIM, TAGLN), and cancer-associated fibroblast markers, without evidence of immune activation. The macrophages were present but appeared to be functioning in a pro-fibrotic rather than immune-activating capacity.
This finding reveals that the location and molecular context of immune cells within a tumor matters as much as their number. Two tumors can have similar macrophage densities yet have completely opposing immune environments depending on the molecular programs those macrophages are running.
Tumor-associated macrophages (TAMs) are emerging as a critical therapeutic target in MSS colorectal cancer. This study suggests that in PD-L1-negative tumors, TAMs occupy a pro-fibrotic, mesenchymal-supportive role that actively contributes to immune exclusion - a state that might be reversible with targeted intervention.
Macrophage repolarization - converting M2-type immunosuppressive macrophages back toward the M1 pro-inflammatory phenotype that promotes anti-tumor immunity - has been explored as a strategy to turn cold tumors hot. This approach could potentially restore CD8+ T cell infiltration and make previously immunotherapy-resistant MSS tumors responsive.
The study also highlights that EMT and immune evasion are deeply linked, potentially through macrophage-mediated mechanisms. Macrophages may play a central role in driving the mesenchymal transition of cancer cells while simultaneously suppressing T cell activity - making them a dual target in MSS CRC.
From a clinical perspective, measuring immune cell PD-L1 by immunohistochemistry could serve as a readily implementable prognostic tool in routine pathology laboratories. Its independence from standard staging parameters could help identify high-risk patients within early-stage disease who might benefit from intensified treatment or closer surveillance.
This study establishes immune cell PD-L1 expression as a robust, independent prognostic biomarker in MSS colorectal cancer - a tumor type that has historically lacked reliable immune markers beyond microsatellite status. The 82% reduction in relapse risk for PD-L1-positive patients is a striking clinical signal.
The integration of bulk RNA sequencing and spatial transcriptomics provides a multi-scale view of CRC biology that goes beyond what conventional tissue analysis can reveal. The spatial data specifically demonstrates that macrophage-enriched niches are the sites where immune-activated and mesenchymal states diverge between PD-L1-positive and PD-L1-negative tumors.
The exploratory nature of the spatial transcriptomics analysis (only four cases) represents a key limitation that must be acknowledged. While the bulk transcriptomic findings are robust, the spatial findings serve primarily as a proof of concept requiring validation in larger, independent cohorts with more samples.
Future studies should test whether combining PD-L1 status with macrophage polarization markers and EMT signatures creates a more powerful prognostic or predictive model, and whether macrophage-targeting therapies can improve outcomes specifically in the PD-L1-negative MSS CRC population - a group with unmet therapeutic need and particularly poor prognosis.