The heterogeneity problem Hepatocellular carcinoma (HCC) is one of the leading causes of cancer-related death worldwide, and it is notoriously heterogeneous - different regions of the same tumor, and different tumors within the same patient, can have entirely distinct genetic and biological profiles.
Why this matters This heterogeneity is a major reason why current treatments fail. A therapy targeting one tumor region may be useless against other regions. Understanding heterogeneity at every level - from DNA to metabolites to immune cells - is essential for designing effective treatments.
Study design Researchers collected 42 samples from eight HCC patients and analyzed them using five cutting-edge technologies simultaneously: whole-exome sequencing, RNA sequencing, mass spectrometry-based proteomics, metabolomics, and CyTOF (single-cell immune profiling). Three independent validation cohorts were also used.
Core conclusion While genomic, transcriptomic, proteomic, and metabolomic heterogeneity was extreme across lesions, immune heterogeneity was comparatively lower and more tractable as a therapeutic target. The team proposed a novel immune-based classification that can guide treatment decisions.
Sample collection Fresh tissue samples (5x5x5 mm each) were collected from 21 lesions including one vascular thrombus across 8 HCC patients. Samples were flash-frozen for sequencing and proteomics, with a separate portion used for CyTOF immune profiling. Peripheral blood and normal liver tissue served as controls.
Genomic and transcriptomic analysis Whole-exome sequencing (WES) was performed on an Illumina HiSeq XTEN platform. RNA sequencing characterized the transcriptome. Single-cell RNA-seq was also performed to resolve cell-level heterogeneity within bulk tumor samples.
Proteomics and metabolomics Quantitative proteomics used multiplexed tandem mass tag (TMT) labeling followed by high-performance liquid chromatography and mass spectrometry. Untargeted metabolomics used liquid chromatography coupled to high-resolution mass spectrometry in both positive and negative ionization modes.
Immune profiling by CyTOF Mass cytometry (CyTOF) used an in-house antibody panel to simultaneously measure over 30 proteins on individual tumor-infiltrating immune cells, enabling precise identification and quantification of all immune cell populations within each tumor sample.
Genomic heterogeneity The study detected 11,266 nonsynonymous DNA mutations, 14,706 RNA-level variants, and 1,875 protein-level variants. Only 20.8% of somatic DNA mutations were expressed at the RNA level, and even fewer (9.32%) were confirmed at the protein level, highlighting the complex filtration of genetic information.
Branch evolution Each tumor showed both shared (trunk) mutations present in all lesions and lesion-specific (branch) mutations. Phylogenetic tree reconstruction confirmed branched tumor evolution, with driver mutations correlating with tumor differentiation status - for example, poorly differentiated lesions carried distinct mutation profiles from well-differentiated ones.
Transcriptome and proteome differences The number of differentially expressed genes varied nearly two-fold between lesions of the same tumor. Only a small subset of these genes was shared across all lesions. Similarly striking differences were found in protein expression and metabolite composition between tumor regions.
Single-cell analysis reveals further heterogeneity Single-cell RNA-seq identified 3-4 distinct HCC cell clusters within each tumor regardless of morphological appearance, and revealed further cell-level heterogeneity not captured by bulk RNA-seq.
Relative immune consistency In contrast to the extreme heterogeneity of tumor cells, the immune status of the tumor microenvironment was comparatively less variable within each patient. This makes the immune environment a more practical target for therapy that aims to work across all lesions.
Three immune subtypes identified CyTOF analysis revealed three distinct HCC immune subtypes: immunocompetent (active immune response), immunodeficient (poor immune infiltration), and immunosuppressive (immune cells present but functionally blocked). Each subtype showed distinct metabolic signatures and cytokine/chemokine expression profiles.
PI3K-Akt and CD8+ T cells PI3K-Akt signaling and ketone body metabolism in tumor cells were associated with the level of CD8+ cytotoxic T cell infiltration - a key measure of anti-tumor immune activity. This suggests metabolic features of tumor cells shape the immune environment.
Key immunosuppressive players The immunosuppressive subtype was characterized by elevated Foxp3+ regulatory T cells (Tregs) and reduced CD45+ immune cell infiltration. The ratio and distribution of these two markers proved sufficient to classify patients into the three immune subtypes using simple immunohistochemistry.
Prognostic value The three-way immunophenotypic classification (immunocompetent, immunodeficient, immunosuppressive) demonstrated significant prognostic value - patients in different immune subtypes had meaningfully different survival outcomes in independent validation cohorts.
Simple clinical implementation The classification can be determined using only two standard immunohistochemical stains - CD45 (total immune cells) and Foxp3 (regulatory T cells) - with established numerical cut-off values. This makes the test feasible in any pathology laboratory worldwide.
Guiding immunotherapy selection Immunocompetent patients may be best candidates for checkpoint inhibitor immunotherapy. Immunosuppressive patients - with high Foxp3+ Tregs blocking immune activity - might benefit from Treg-targeting strategies. Immunodeficient patients may need approaches to increase immune cell recruitment before immunotherapy will work.
Addressing the treatment gap Currently, sorafenib and checkpoint inhibitors show limited efficacy in HCC. The novel classification provides a path to match the right immunotherapy strategy to the right patient, potentially transforming treatment outcomes for this difficult-to-treat cancer.
Small primary cohort The multiomic analysis was performed on only 8 patients, which limits statistical power and may not fully capture the diversity of HCC worldwide. The findings were validated in three independent cohorts, but the primary discovery cohort remains small.
Causality vs. correlation While associations between metabolic pathways, immune infiltration, and tumor subtypes were identified, the causal mechanisms linking tumor metabolism to immune microenvironment composition remain to be experimentally validated.
Translational gap Previous genomic studies of HCC heterogeneity have not yet translated into improved therapies. Whether the immunophenotypic classification can prospectively improve treatment selection in clinical trials remains to be demonstrated.
Future directions Prospective clinical trials using this two-marker classification to stratify patients for immunotherapy, combined with longitudinal monitoring of immune subtype stability, will be key next steps toward implementing these findings in clinical practice.