Colorectal cancer (CRC) is the third most common cancer worldwide and the second leading cause of cancer-related mortality. While early-stage CRC patients generally exhibit favorable overall survival (OS) rates, the prognosis for metastatic CRC (mCRC) remains poor, with a survival rate < 15%. Targeted combination therapy remains the main treatment strategy for mCRC, with a median OS (mOS) of only 25–30 months. With the widespread use of immune checkpoint inhibitors (ICIs), significant progress has been made in microsatellite instability-high (MSI-H) mCRC. However, patients with deficient mismatch repair (dMMR)/MSI-H only account for 5% of all CRC patients and 95% of microsatellite stable (MSS) mCRC patients are not recommended to receive direct immunotherapy. Given the heterogeneity of tumors, MSI status alone is insufficient to accurately predict the efficacy of immunotherapy in CRC. Specifically, some MSI-H CRC patients may exhibit primary resistance to treatment, while some MSS CRC patients could potentially benefit from immunotherapy. Therefore, there is an urgent need to identify new immunotherapy biomarkers to enhance the precision of immunotherapy for CRC.
Immune microenvironment characteristics across different pathologic CRC subtypes
CRC predominantly consists of adenocarcinomas, although other more rare histologic subtypes, such as mucinous adenocarcinoma (MC), signet-ring cell carcinoma (SRCC), adenosquamous carcinoma (ASC), and medullary carcinoma (MeC), also exist. These subtypes are distinguished by unique histopathologic features and distinct immune microenvironment profiles. These differences will likely affect the efficacy of immunotherapy but this issue has received little attention in previous studies.
MC
MC accounts for approximately 10% of CRC cases. MC is characterized by abundant mucin production and the presence of mucin pools, which are believed to have a critical role in tumor progression, invasion, survival, and evasion of host immune responses. Notably, some studies have reported that MC exhibits increased lymphocyte infiltration, enhanced programmed death-ligand-1 (PD-L1) expression, and a higher frequency of MSI compared to non-MC. These features may be linked to Lynch syndrome and the CpG island methylator phenotype (CIMP), suggesting a more immunogenic tumor microenvironment (TME) in a subset of these cases1.
SRCC
SRCC is a highly aggressive CRC subtype with a poor prognosis. Studies have shown that hypermethylated SRCC is characterized by a high tumor mutation burden (TMB), MSI-H, elevated PD-L1 expression, and increased immune cell infiltration, such as CD8+ T cells, which suggests that this subtype may be sensitive to immunotherapy. However, current clinical observations have shown that SRCC patients exhibit dismal clinical outcomes and derive limited benefit from immunotherapy with the underlying mechanisms requiring further investigation2.
ASC
ASC is an uncommon CRC subtype that combines glandular and squamous PD-L1 histologic features. The ASC immune microenvironment exhibits a dual nature with a pronounced immunogenicity counterbalanced by substantial immunosuppressive elements. High expression of immune checkpoint molecules, such as is frequently observed in ASC, contributes to establishing an immunosuppressive microenvironment that limits immune response efficacy. Moreover, most reported cases of ASCs are MSS mCRCs, which may partly explain the poor response to immunotherapy3.
Serrated adenocarcinoma (SAC)
SAC, a variant of colorectal adenocarcinoma by the World Health Organization (WHO), is considered the ultimate progression of serrated lesions. The SAC immune microenvironment exhibits distinct characteristics. Current research suggests that the SAC TME may impair recruitment or survival of cytotoxic T lymphocytes, potentially leading to a relatively weakened antitumor immune response. Reduced expression of β2-microglobulin (B2M) in SAC tumor cells, which results in diminished tumor neoantigen presentation, has been identified as a key mechanism contributing to the formation of an immunosuppressive microenvironment in SACs4.
MeC
MeC is a rare, poorly differentiated CRC subtype that is often misdiagnosed but with a relatively good prognosis. MeCs have a unique immune microenvironment with extensive lymphocyte infiltration and high interferon-gamma (IFN-γ) pathway activation. Analysis of 47 MeC patients showed a high prevalence of dMMR/MSI-H, ARID1A mutations and ASCL2 gene amplification. Genes related to T-cell activation were upregulated, while oncogenic pathways were downregulated. There was abundant immune cell infiltration, especially clonally expanded CD8+ T cells. Both MSI-H and MSS MeC had more immune cell infiltration than MSI-H non-MeC. In a retrospective study involving 47 MeC patients treated with ICIs, 30 had a partial or complete response and 10 had stable disease. The median progression-free survival was 14.33 months and the mOS was 38.33 months, which was far better than non-MeC5. There was no significant response difference between MSI-H and MSS MeCs, suggesting that even MSS MeCs (traditionally “cold”) benefit from immunotherapy due to the active immune microenvironment. MeC is a “hot” CRC subtype that is highly responsive to immunotherapy independent of MSI status. This feature shows the potential for CRC immunotherapy and the need to determine its immunogenicity and improve the diagnosis and applications.
Immunotherapy biomarkers in CRC
Recent research on biomarkers that predict immunotherapy efficacy in CRC is summarized in Table 1. These biomarkers mainly focus on the mutation status of specific genes or the composition and characterization of tumor immune cells [e.g., tumor infiltrating lymphocytes (TILs)6] The application of these biomarkers is expected to improve the precision of immunotherapy for CRC.
Current research on biomarkers for precision immunotherapy in CRC
TMB
The TMB measures the total mutations per megabase in tumor cells with a high TMB linked to increased neoantigen production and enhancing immune recognition. The TMB is a potential biomarker for predicting immunotherapy response across cancers, including mCRC. Recent studies highlight the TMB predictive value. Tremelimumab combined with durvalumab was used as a last-line treatment for patients with refractory CRC in the CTGCO.26 trial. The TMB status was assessed through plasma testing. Patients with TMB > 28 mutations/Mb could benefit from immunotherapy7. Further analysis from the KEYNOTE 177 clinical trial revealed that, even for MSI-H CRC, higher TMB levels often indicate higher responsiveness to immune checkpoint blockade (ICB) therapy8. Therefore, the TMB is a marker of great clinical value for CRC. The only remaining challenge is to determine a more accurate cut-off value.
Polymerase epsilon (POLE)/polymerase delta 1 (POLD1) mutations
DNA POLE subunit 1 and POLD1 mutations are considered significant predictive biomarkers for immunotherapy efficacy in MSS CRCs. DNA POLE subunit 1 and POLD1 mutations are frequently associated with a high TMB. POLE-mutated CRCs (approximately 1% of CRCs) exhibit enhanced tumor immunogenicity that is characterized by increased TIL levels, upregulated PD-L1 expression, and elevated cytotoxic T-cell markers and effector cytokines, which results in durable clinical benefits from ICB therapy9. POLD1 mutations, which have a critical role in cell cycle regulation and DNA damage repair (DDR), are often associated with microsatellite instability, suggesting that patients with POLD1-mutated CRCs may also benefit from immunotherapy. Currently, POLE and POLD1 mutations are recommended as markers for predicting benefit from immunotherapy in patients with MSS CRCs6.
Epigenetic gene mutations
Mutations in epigenetic regulation genes have been increasingly shown to be closely associated with immunotherapy. For example, ARID1A mutations define an immunologically active subgroup in MSS CRCs that is characterized by a higher TMB, increased frameshift mutations, and enhanced immune responses, such as T-cell infiltration and IFN-γ pathway activation. These traits suggest that MSS CRC patients with ARID1A mutations may respond better to immunotherapy10. Another study showed that KMT2C/D loss-of-function mutations are associated with a higher TMB, enhanced immune responses (e.g., increased PD-L1 expression and CD8+ T cell infiltration), and improved efficacy of programmed death-1 (PD-1)/PD-L1 therapy in CRC, suggesting the potential as predictive biomarkers for immunotherapy11.
DDR mutations
Genes involved in DDR, including ATM, BRCA2, ERCC2/4, FANCA, and CHEK1/2, and MMR-related genes, such as MLH1, MSH2, and MSH6, are integral to maintaining genomic integrity. Mutations in these genes can promote the generation of neoantigens, thereby increasing tumor immunogenicity and enhancing the TMB. For instance, DDR mutations have been linked to the development of hypermutator phenotypes. Studies suggest that MSS-CRC patients harboring DDR mutations exhibit stronger immune responses and greater sensitivity to ICIs compared to MSS-CRC patients without such mutations. Nevertheless, the pathogenicity and prevalence of specific DDR mutations in MSS-CRCs remain poorly defined and are significantly lower than cancers, such as endometrial carcinoma11.
Immunoscore
Studies have shown that the immunoscore, by quantifying the density of CD3+ and CD8+ T-cells in tumors and invasive margins, effectively predicts the risk of recurrence and survival rates in CRC patients. A high immunoscore is significantly associated with a lower recurrence risk, higher disease-free survival, and overall survival. The predictive ability of a high immunoscore is independent of TNM staging and MSI status. As a reflection of the antitumor immune microenvironment, the immunoscore can assist in identifying patients suitable for immunotherapy12. A study from the 2024 European Society for Medical Oncology (ESMO) conference reported that chemotherapy combined with targeted and immune treatment for patients with proficient mismatch repair (pMMR)/MSS CRC and high immune scores achieved an objective response rate (ORR) of 74% and a disease control rate (DCR) of 100%.
PD-L1 expression
While high PD-L1 expression is often linked to better responses to ICB therapy in some tumor types, such as non-small cell lung cancer (NSCLC) and gastric cancer, the predictive value in CRC, especially in MSS patients, is not satisfactory. PD-L1 expression in CRC was shown not to be associated with response or survival in the registration studies9. The challenges in using PD-L1 as a CRC biomarker include sample heterogeneity, risk of false negatives, and the dynamic nature of PD-L1 expression influenced by the TME. Further research is needed to determine the predictive value of PD-L1 expression in immunotherapy for CRC, as well as to investigate the potential of combining PD-L1 with other biomarkers, such as immune scores, TMB, CD4+ T cell frequency, and the CD4+/CD8+ ratio.
B2M and Janus kinases 1 and 2 (JAK1/2) mutations
B2M has a crucial role in tumor antigen presentation and mutations in JAK1/2 can impair the transmission of signals through the interferon pathway. Typically, mutations in B2M and JAK1/2 are associated with resistance to ICB therapy, as occurs in malignant melanoma. However, an apparently paradoxical finding has emerged in patients with MSI-H CRC. A post-hoc analysis of the KEYNOTE-177 study cohort revealed that patients with B2M mutations or loss of expression could achieve sustained therapeutic responses to ICB therapy8. Similarly, a study conducted in China reported comparable results. Notably, MSI-H CRC patients with JAK1/2 mutations exhibit superior responses to anti-PD-1 therapy compared to MSI-H CRC patients without such mutations13. Researchers believe that these discrepant findings of CRC with B2M or JAK1/2 mutations are mainly due to changes in the immune microenvironment and the TMB. In MC38 B2M-knockout tumors, the number of type 1 conventional DCs (cDCs1) decreases while type 2 conventional DCs (cDCs2) increases. cDCs2 partially compensate for the loss of cDC1 function by activating CD4+ T cells14. Additionally, CD4+ T cells limit the differentiation and proliferation of macrophages into an immunosuppressive M2 state, thereby promoting the anti-tumor immune response. Moreover, MSI-H CRC patients carrying B2M or JAK1/2 mutations have a higher TMB13, which implies that tumor cells may generate more tumor neoantigens, making the tumor cells more easily recognized and attacked by the immune system. Deeper mechanisms deserve further exploration.
CD8+ MeTIL score
DNA methylation detection is an important method for the molecular diagnosis of tumors, especially in the early diagnosis of cancer, but it also has great potential in predicting the effect of antitumor treatment. A study analyzed the genome-wide DNA methylation profiles of immune cells and colonic epithelial cells to identify CD8+ T cell-specific differentially methylated positions (DMPs) and developed the CD8+ MeTIL score to assess CD8+ TILs and survival outcomes in CRC. The results showed that the DNA methylation signature for the CD8+ TIL (CD8+ MeTIL) score was significantly lower in the MSI-H group than the microsatellite instability-low (MSI-L)/MSS group and patients with abundant CD8+ TILs (low CD8+ MeTIL score) had the best OS in both MSI-H and MSS cohorts. The CD8+ MeTIL signature has the potential to be a useful biomarker, including immunotherapy15.
Gene expression profiles (GEPs)
GEP is often used in tumor immunology research to describe the expression characteristics of gene sets associated with T-cell inflammation in the TME, particularly genes related to tumor antigen presentation, cytokine signaling, and adaptive immune responses. A study analyzing the immune GEP of patients with NSCLC, head and neck squamous cell carcinoma (HNSCC), and melanoma (SKCM) following PD-1 blockade therapy reported that immune GEP was associated with treatment efficacy16. An evaluation of patient samples from 22 tumor types across 4 KEYNOTE clinical trials revealed that the T-cell-inflamed GEP can independently predict responses to pembrolizumab and holds promise for guiding precision immunotherapy. For example, it was noted that some MSI-H CRC patients with low GEPs may be associated with ineffective immunotherapy and while most MSS CRCs have a low GEP, a subset of high-GEP patients may still benefit from immunotherapy17.
Precision immunotherapy in CRC
Precision immunotherapy of CRC hinges on in-depth research and application of immunotherapy biomarkers. ICIs are highly effective in MSI-H CRC patients, yet this subgroup accounts for only approximately 5% of CRC cases, making it crucial to expand immunotherapy applicability. Recent studies have uncovered emerging biomarkers, such as TMB, POLE/POLD1 mutations, epigenetic changes (e.g., ARID1A and KMT2D), and DDR gene mutations, therefore presenting new treatment chances for MSS patients. Notably, MSS patients with a high TMB or POLE/POLD1 mutations respond well to immunotherapy. Moreover, the CRC pathologic heterogeneity and complex immune microenvironment pose challenges. MeC, for example, has unique immune activation features, such as lymphocytic infiltration, elevated IFN-γ pathway activity, increased MMR protein deficiency, and a 10% POLE mutation rate, making MeC highly responsive to immunotherapy. This immune-active phenotype exhibits strong antitumor immunity and immunotherapeutic efficacy independent of MSI status (Figure 1). Beyond focusing on the biomarkers’ guiding implications, the correlations with other elements in the TME should be studied. Whether biomarkers cooperate with cytokines, such as tumor necrosis factor-α, during the immune response18 and interfere with the epithelial-mesenchymal transition19 to affect tumor recurrence and metastasis should be determined. In addition, whether biomarkers can influence efficacy through associations with the gut microbiota and microbial metabolic profile should also be determined20.
MeC exhibits unique immune activation characteristics independent of MSI status. MeC, medullary carcinoma; MMR, mismatch repair; TILs, tumor-infiltrating lymphocytes; MSI-H, microsatellite instability-high; MSS, microsatellite stable.
In addition, advanced detection technologies, including single-cell sequencing, spatial transcriptomics, dynamic liquid biopsy, immune repertoire sequencing, and multi-omics integrated analysis, offer significant advantages. These technologies can effectively address issues related to sample heterogeneity, which enables precise identification of the functional states of key cell subsets. Moreover, the technologies allow for real-time monitoring of marker dynamic changes, which is crucial for early prediction of immunotherapy responses and drug resistance. Ultimately, these capabilities provide a robust scientific basis for personalized treatment and precision immunotherapy. When leveraging these markers to direct precision immunotherapy, it is of utmost importance to comprehensively acknowledge that the therapeutic implications of the same biomarker can exhibit substantial disparities among diverse tumor types. For example, high PD-L1 expression generally indicates a favorable response to immunotherapy in NSCLC and gastric cancer, while in CRC high PD-L1 expression needs to be comprehensively interpreted in combination with the MSI status. B2M and JAK1/2 mutations are associated with a better prognosis in CRC and are closely linked to immunotherapy resistance in melanoma. Similarly, colorectal MeC has a relatively good prognosis, but in thyroid cancer MeC represents the subtype with the highest degree of malignancy. Therefore, in clinical research these markers should be gradually incorporated into the process of protocol design and implementation. Through combined and stratified analysis, the populations with advantages for immunotherapy can be screened out to pursue a higher treatment efficiency. This approach not only contributes to enhancing the clinical accessibility of biomarkers but also further promotes the application and development of precision immunotherapy in the clinical practice of CRC.
Conflicts of interest statement
No potential conflicts of interest are disclosed.
Author contributions
Conceived and designed the analysis: Yanqiao Zhang, Chao Liu, Ya Lan, Hong Wang.
Wrote the paper: Chao Liu and Ya Lan.
Footnotes
↵*Chao Liu and Ya Lan contributed equally to this work.
- Received January 16, 2025.
- Accepted February 20, 2025.
- Copyright: © 2025 The Authors
This work is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.