Colon adenocarcinoma (COAD) is one of the most common and deadly cancers worldwide, with approximately 1.93 million new cases and over 900,000 deaths reported in 2023. While treatments like surgery, chemotherapy, and targeted therapies have improved outcomes for early-stage patients, those with advanced disease still face poor survival because of tumor heterogeneity and resistance to treatment.
Two important biological processes have been studied separately in colon cancer: programmed cell death (PCD), which is the body's natural mechanism for eliminating damaged or dangerous cells, and N6-methyladenosine (m6A) modification, a chemical change to RNA molecules that controls how genes are read and expressed. When either process goes wrong, cancer cells can survive and grow unchecked.
m6A modification works through three types of proteins: writers (such as METTL3), which add the chemical tag to RNA; readers (such as YTHDC1), which recognize the tag and act on the modified RNA; and erasers, which remove it. This system influences how stable a messenger RNA molecule is and how efficiently it is translated into protein. In cancer, these processes are frequently altered.
Previous research had studied m6A modification and programmed cell death independently, leaving a gap in understanding how the two interact. This study aimed to bridge that gap by building an integrated model that combines both layers of biology to better predict patient outcomes and identify new therapeutic targets in colon adenocarcinoma.
Researchers analyzed 1,379 genes spanning 14 different programmed cell death pathways using data from The Cancer Genome Atlas (TCGA) colon adenocarcinoma dataset. They identified which genes were differentially expressed between tumor and normal tissue, which were associated with patient survival, and which correlated with known m6A regulatory enzymes. This three-way intersection produced 31 hub genes that were prognostically relevant and linked to m6A regulation.
A statistical technique called LASSO Cox regression was then used to narrow these 31 genes down to a final 21-gene core signature, named the m6A-PCD Integrated Signature (MCDI). Each patient received a composite MCDI risk score calculated from the expression levels of these 21 genes, with higher scores indicating a greater predicted risk of death.
To validate the MCDI, the model was tested in four independent patient datasets: the TCGA-COAD cohort for internal validation, and three external datasets (GSE39582, GSE33113, and GSE38832). In all cohorts, patients with high MCDI scores had significantly shorter overall survival. The area under the ROC curve for predicting 1-year survival reached 0.783 to 0.805 in external cohorts, outperforming previously published single-mechanism models.
Clinical tissue samples from 10 colon cancer patients at Fujian Medical University Union Hospital were also collected and used to validate key gene expression findings in real patient specimens. Single-cell RNA sequencing data from publicly available colon cancer datasets (GSE132465 and GSE205506) were used to determine which cell types within tumors expressed the key signature genes.
Analysis confirmed that PCD-related genes are extensively dysregulated in colon tumor tissue, with 438 genes significantly upregulated and 393 downregulated compared to adjacent normal tissue. Among the 21 final MCDI genes, 15 were upregulated and 6 were downregulated in tumors. Multivariate Cox regression identified five of these as independent prognostic factors: MIR210, STK25, TGFB2, TRIM6, and TRIM68.
MCDI scores increased progressively with tumor stage. Stage IV tumors showed significantly higher scores than Stage I and II, and patients with lymph node involvement (N1 or N2) had higher scores than those without. Patients who died during the study period had substantially higher MCDI scores than those who remained alive, confirming the model's clinical relevance.
The immune microenvironment of high-MCDI patients was found to be significantly more suppressed. Compared to low-risk patients, the high-risk group had higher levels of M0 macrophages, regulatory T cells (Tregs), and plasma cells, all of which can suppress effective immune responses against cancer, along with reduced monocyte infiltration. TIDE analysis further confirmed a greater immune evasion potential in high-MCDI tumors.
Interestingly, MCDI scores were not significantly associated with tumor mutational burden (TMB) or microsatellite instability (MSI) status, suggesting that the prognostic and immunological features captured by MCDI operate through a distinct biological mechanism that is independent of these commonly used immunotherapy biomarkers.
Among the five independent prognostic genes, STK25 (serine/threonine kinase 25) emerged as the most clinically interesting. Single-cell RNA sequencing revealed that STK25 is specifically and markedly upregulated in malignant epithelial cells within colon tumors, but not in the same epithelial cells from adjacent normal tissue. This tumor-specific expression pattern was confirmed in two independent single-cell datasets.
To confirm these findings in human tissue, the researchers tested STK25 protein and mRNA levels in 10 matched pairs of colon cancer tissue and adjacent normal tissue. Both qPCR and western blot analyses consistently showed substantially higher STK25 expression in the tumor samples. Patients with high STK25 expression had significantly shorter overall survival on Kaplan-Meier analysis, particularly in association with lymph node metastasis.
To understand what STK25 actually does in cancer cells, researchers used a technique called siRNA knockdown to silence the gene in two colon cancer cell lines (RKO and LOVO). When STK25 was silenced, the cancer cells underwent significantly more apoptosis (programmed cell death) as measured by flow cytometry. This demonstrates that STK25 actively helps colon cancer cells avoid dying, functioning as a survival factor for tumor cells.
These findings position STK25 as both a reliable prognostic marker and a potential therapeutic target. Its high and specific expression in tumor epithelial cells, combined with the functional evidence that its loss triggers cancer cell death, makes it an attractive candidate for future drug development aimed at colon adenocarcinoma.
Researchers next asked what drives the high expression of STK25 in colon cancer cells. Correlation analysis of 23 known m6A regulatory enzymes against STK25 expression identified that METTL3 (an m6A writer) and YTHDC1 (an m6A reader) showed the strongest positive correlation with STK25 levels. The RM2Target database confirmed these two proteins as the highest-confidence predicted regulators of STK25.
When METTL3 or YTHDC1 were silenced using siRNA in cancer cell lines, STK25 mRNA levels dropped significantly. To confirm direct interaction, RNA immunoprecipitation (RIP) assays were performed, showing that both METTL3 and YTHDC1 proteins physically bind to STK25 messenger RNA in cancer cells. This provides strong evidence for a direct regulatory relationship.
MeRIP-qPCR analysis confirmed that METTL3 adds m6A chemical modifications specifically to the STK25 mRNA molecule. When METTL3 was knocked down, the m6A modification level on STK25 mRNA decreased substantially. This means that METTL3 installs the m6A tag, YTHDC1 reads it and stabilizes the mRNA, and the resulting stable STK25 protein then suppresses apoptosis in cancer cells.
Taken together, these experiments define a novel regulatory chain: METTL3 methylates STK25 mRNA, YTHDC1 reads that modification to protect the mRNA from degradation, and the resulting high STK25 protein levels suppress cancer cell death. Disrupting any step in this chain could represent a therapeutic strategy to increase apoptosis in colon tumor cells.
Because the MCDI signature is enriched with immune-related biological pathways, researchers tested whether the score could predict how well patients respond to immune checkpoint blockade (ICB) therapy, which uses drugs to reactivate the immune system against cancer. The MCDI was applied to two independent immunotherapy datasets from renal cell carcinoma and melanoma patients treated with anti-PD-L1 drugs.
In both datasets, patients with low MCDI scores responded significantly better to anti-PD-L1 treatment than those with high MCDI scores. This suggests that tumors classified as low-risk by MCDI may have a more immunologically active environment that can be further activated by checkpoint drugs, while high-MCDI tumors may be inherently immunosuppressed and resistant to this approach.
A prognostic nomogram was constructed combining the MCDI score with patient age and T stage (depth of tumor invasion). This clinical tool, which translates the mathematical model into a simple point system, allows physicians to estimate individual probabilities of 1-year and 3-year survival for each patient, potentially guiding treatment decisions in clinical practice.
While the MCDI's immunotherapy predictive value requires further validation specifically in colon cancer immunotherapy cohorts, it offers a complementary approach to existing biomarkers like TMB and MSI, capturing biological information these markers do not reflect. The model's independent prognostic value across multiple clinical strata suggests it could support more personalized treatment planning in COAD.
This study developed and validated the MCDI, a 21-gene prognostic model that integrates two previously separate dimensions of colon cancer biology: m6A RNA modification and programmed cell death. By combining these layers, the model achieves better predictive performance across multiple independent patient cohorts than prior single-pathway models.
The discovery of the METTL3-YTHDC1-STK25 regulatory axis represents a mechanistic advance beyond simply identifying correlated markers. By showing that m6A modification directly stabilizes STK25 mRNA and thereby suppresses apoptosis, the research provides a biologically grounded explanation for STK25's role as a prognostic risk factor and a rationale for targeting this axis therapeutically.
The study acknowledges important limitations: the patient cohorts used for model building are relatively small, and prospective multicenter validation is needed. The mechanistic work on METTL3 and YTHDC1 regulation of STK25 would benefit from additional confirmation experiments. These are recognized directions for future research rather than flaws that undermine the current findings.
Looking ahead, the researchers propose that the MCDI framework could support clinical translation in two ways: as a risk stratification tool for identifying patients most likely to benefit from immunotherapy, and as a roadmap for drug development targeting the m6A-STK25 axis to enhance apoptosis in resistant colon cancer cells. Validation in dedicated COAD immunotherapy cohorts is the next critical step.