Bile Acid Subtypes and Immune Dysfunction in CRC

Colorectal cancer (CRC) is biologically heterogeneous, and patients with similar clinicopathological classifications can show very different immune states, outcomes, and responses to immune checkpoint inhibitors. In the open-access study published on 12 January 2026, Feng et al. examined whether bile acid metabolism could provide a biologically informative basis for molecular subtyping. The authors linked transcriptomic patterns in TCGA-COAD to survival, immune-cell infiltration, candidate hub genes, and computational estimates of immunotherapy relevance in the reference study.

Study Background and Research Question

Bile acids are not only digestive molecules; they also function as signaling metabolites in the gastrointestinal tract. Dysregulated bile acid exposure has been associated with intestinal inflammation, epithelial stress, and CRC progression. However, the relationship between bile acid-associated transcriptional programs and the tumor immune microenvironment (TIME) remains incompletely defined.

This question is clinically relevant because immune checkpoint inhibitors have changed treatment for selected patients, particularly those with high microsatellite instability, while primary resistance remains a major obstacle. A molecular classification that captures both metabolic state and immune context could help explain why tumors with apparently similar diagnoses behave differently.

The central research question was therefore whether expression patterns related to bile acid metabolism could divide colon adenocarcinoma into clinically and immunologically distinct groups. The investigators also asked whether genes associated with these groups could serve as reproducible markers of tumor biology, prognosis, or predicted immunotherapy sensitivity.

Key Innovation from the Reference Study

The main innovation was to use bile acid metabolism as an organizing principle for integrative CRC subtyping rather than treating metabolism and immunity as separate analytical domains. Using unsupervised consensus clustering, the authors classified patients into bile-high and bile-low molecular groups. They then connected this classification with overall survival, immune infiltration, differential gene expression, protein–protein interaction analysis, and Cox regression.

This design is useful because it moves beyond a single-gene association. A metabolic subtype can represent a broader transcriptional state, while the downstream hub-gene analysis can identify a smaller set of candidates for clinical or experimental validation. The study ultimately highlighted CLCA1, UGT2A3, and ZG16 as the three principal genes associated with the bile acid-related phenotype.

The innovation should nevertheless be interpreted as a discovery and stratification framework, not as proof that these genes directly control bile acid metabolism or immune resistance. The analyses identify coordinated associations in patient-derived data. Functional experiments would be needed to establish causality.

Methods and Experimental Design Insights

The discovery analysis used transcriptome and clinical data from The Cancer Genome Atlas colon adenocarcinoma cohort. Patients were grouped according to the expression behavior of bile acid metabolism-related genes through unsupervised consensus clustering. This approach avoids imposing predefined classes, but it also makes the resulting subtype structure dependent on the selected gene set, normalization strategy, and clustering parameters.

After subtype assignment, the researchers compared overall survival and immune-cell infiltration between groups. Differentially expressed genes were then examined to identify transcripts associated with the bile-high and bile-low states. A protein–protein interaction network helped prioritize connected candidate genes, and Cox proportional hazards regression was used to evaluate their relationship with survival.

Validation was performed in a Gene Expression Omnibus cohort and in independent clinical samples. This external checking step is important because genes identified in a single public dataset may reflect cohort-specific technical or clinical features. The study also assessed relationships between the three hub genes and tumor immune dysfunction using the Tumor Immune Dysfunction and Exclusion, or TIDE, score.

Protocol Parameters

• Discovery cohort: use TCGA-COAD transcriptome and clinical data to establish the bile acid metabolism-associated molecular groups, as performed in the reference study.
• Subtype analysis: apply unsupervised consensus clustering to the selected bile acid metabolism-related expression features; clustering settings should be documented because they can influence class stability.
• Clinical comparison: compare overall survival and immune-cell infiltration between the bile-high and bile-low groups using prespecified statistical procedures.
• Candidate prioritization: integrate differentially expressed genes with a protein–protein interaction network and Cox regression rather than selecting markers from fold change alone.
• Independent validation: test CLCA1, UGT2A3, and ZG16 in an external GEO dataset and independent clinical material before interpreting them as broadly transferable biomarkers.
• Wet-lab follow-up: for qPCR confirmation, use matched RNA inputs, include genomic DNA contamination removal, and define reference genes and technical replicates before testing whether transcript-level subtype differences are reproducible.

The first five parameters describe the published analytical design. The final point is a workflow recommendation for researchers translating the computational signature into targeted gene-expression measurements; it is not a parameter reported by Feng et al.

Core Findings and Why They Matter

The bile-low group had significantly shorter overall survival than the bile-high group, with a reported p value of 0.0049 in the reference study. This result supports the idea that reduced activity of the bile acid-associated transcriptional program marks a clinically unfavorable state in the analyzed cohort. It does not establish that low bile acid metabolism is independently prognostic after adjustment for every relevant clinical variable.

The immune profile was more complex than a simple high-inflammation versus low-inflammation model. CD8-positive T-cell infiltration was higher in the bile-low group, with p less than 0.05, and M1 macrophage infiltration was also higher, with p less than 0.01. Because the bile-low group nevertheless had poorer survival, immune-cell abundance alone did not correspond to a favorable outcome in this dataset. The finding raises the possibility that the functional state, spatial distribution, or coordination of infiltrating cells may matter more than cell counts alone, although those explanations were not directly tested.

CLCA1, UGT2A3, and ZG16 were all downregulated in tumor tissues across the TCGA-COAD and GEO datasets and in the independent clinical samples described by the authors. Among the three, high CLCA1 expression showed a significant association with favorable overall survival, with p less than 0.001. UGT2A3 and ZG16 showed weaker survival associations that did not reach statistical significance in the reported analysis, with p values of 0.23 and 0.17, respectively.

These results argue for a distinction between a multi-gene biological signature and an individual prognostic marker. The three genes may jointly characterize a bile acid-related epithelial or immune context, whereas CLCA1 appears to have the clearest single-gene survival signal in this study. UGT2A3 and ZG16 should therefore be regarded as components of the candidate signature rather than independently validated survival predictors.

All three genes were negatively correlated with TIDE score: CLCA1 showed R = −0.24, UGT2A3 showed R = −0.15, and ZG16 showed R = −0.14, with the associations reported as statistically significant in the study. These correlations suggest that higher expression of the genes is linked to a more favorable computational immune-response profile. However, TIDE is an in silico predictor and cannot substitute for measured response to immune checkpoint therapy in a treated clinical cohort.

Comparison with Existing Internal Articles

The internal article Bile Acid Metabolism Subtypes Define CRC Immune Markers and Prognosis presents the same study from a concise biomarker-stratification perspective. Its emphasis on CLCA1, UGT2A3, and ZG16 is consistent with the reference paper, but the primary publication provides the essential methodological context: subtype construction, survival comparison, immune-infiltration analysis, PPI prioritization, Cox regression, and external validation.

That distinction matters for literature interpretation. A short summary can efficiently communicate the main result, but it should not be treated as independent replication. The strongest contribution of Feng et al. is the integration of multiple analytical layers around bile acid metabolism, while the main unresolved issue is whether the signature remains predictive in prospectively collected, treatment-annotated cohorts.

Limitations and Transferability

Several limitations constrain how the findings should be used. First, the discovery work is retrospective and relies heavily on public transcriptomic data. Such datasets are valuable for hypothesis generation, but they can contain batch effects, incomplete clinical annotation, and differences in tissue composition. External GEO validation improves confidence in reproducibility but does not remove these concerns.

Second, consensus clustering identifies patterns within the available expression matrix; it does not prove that the resulting bile-high and bile-low groups represent discrete biological states in every CRC population. The analysis focused on colon adenocarcinoma, so transfer to rectal cancer, metastatic lesions, or other gastrointestinal tumors should not be assumed without dedicated testing.

Third, transcript abundance is not equivalent to bile acid concentration, enzymatic activity, or pathway flux. Direct metabolomic measurements and spatially resolved analyses would be needed to determine whether the expression-defined subtypes correspond to distinct bile acid environments in tumors. Similarly, the observed differences in CD8-positive T cells and M1 macrophages do not establish how these cells function within the tumor.

Fourth, the study did not demonstrate that CLCA1, UGT2A3, or ZG16 causes immune dysfunction, treatment resistance, or poor prognosis. The TIDE correlations are hypothesis-generating, and the survival results should not be interpreted as evidence that the markers predict actual immune checkpoint inhibitor benefit. Functional perturbation, paired tumor profiling, and prospective response cohorts would be appropriate next steps.

For researchers, the most defensible current use of the work is as a candidate framework for stratification and validation. The three-gene panel can be tested in independent cohorts, but its clinical value should be assessed alongside established molecular features, treatment history, tumor purity, microsatellite status, and other relevant covariates.

Research Support Resources

Why this cross-domain matters, maturity, and limitations

Connecting a transcriptomic CRC study with targeted reverse-transcription and qPCR workflows can help determine whether computationally identified markers are measurable in local biospecimens. This bridge is useful because the reference study depends on expression data, whereas translational laboratories may need a smaller assay for tissue, cell, or low-input RNA samples. It remains an early validation step: qPCR can confirm relative transcript abundance, but it cannot by itself reproduce the full subtype model, quantify bile acid metabolism, or establish treatment response.

Practical qPCR support

For researchers adapting the CLCA1, UGT2A3, and ZG16 findings to gene expression analysis by qPCR, the HyperScript™ III RT SuperMix for qPCR (with gDNA wiper) (SKU K1585) is designed for two-step qRT-PCR workflows. Its HyperScript III Reverse Transcriptase is described as having enhanced thermal stability, reduced RNase H activity, and improved fidelity, with intended use in applications including reverse transcription of low-concentration RNA and high-GC content RNA reverse transcription. The included gDNA wiper supports genomic DNA contamination removal before cDNA synthesis, while the resulting products are compatible with qPCR reagent formats for SYBR Green and probe-based assays. These features may support technical follow-up of the paper’s candidate genes, but biological interpretation still depends on appropriate controls, normalization, independent samples, and validation against the original study design.