Hepatocellular carcinoma (HCC) is a malignant tumor arising from hepatocytes and is characterized by marked heterogeneity in histological features, molecular alterations, and clinical behavior. Due to variations in tumor etiology, genetic alterations, immune microenvironment, imaging phenotypes, and treatment responses across patients, HCC is increasingly recognized as a group of molecularly distinct subtypes rather than a single disease entity. In line with this evolving perspective, multiple transcriptome-based molecular classification systems have been developed [
1,
2]. However, most of these classification systems are grounded in gene expression profiles or mutational patterns and do not adequately reflect the functional state of tumor cells, particularly in terms of preserved liver-specific functions or shifts in metabolic profiles. Clinically, patients present wide-ranging imaging features, tumor growth rates, and responses to immunotherapy [
3]. Yet current classification systems are often insufficient to fully explain for or predict these clinical variations. Adding to the complexity, some HCCs share biological characteristics with hepatocellular adenoma (HCA) or exhibit morphological overlap with cholangiocarcinoma, complicating the distinction of their biological behavior based solely on histopathological diagnosis. These challenges highlight the need for a more refined classification system that considers the degree of functional dedifferentiation, the preservation of liver-specific metabolic programs, and similarities in developmental lineage [
4,
5].
In this issue, Aoki et al. [
6] propose a novel molecular framework for classifying HCC that integrates compartmentalized hepatic metabolic functions with key oncogenic signaling pathways. A central tenet of this framework is the mutual exclusivity between preserved liver-specific metabolic functions—such as bile acid, fatty acid, ammonia, and drug metabolism—and glycolysis activation. This metabolic shift is not merely an energy source substitution but is closely associated with cellular dedifferentiation, spatial hepatocellular identity loss, and more aggressive tumor phenotypes.
Using transcriptomic data from five large cohorts, the researchers define five molecular HCC subtypes according to dominant metabolic and signaling features: (1) a Wnt/β-catenin-high subtype retaining hepatocellular functions; (2) an interleukin 6-Janus kinase-signal transducer and activator of transcription 3 (IL6-JAK-STAT3)-high subtype resembling inflammatory HCA; (3) a glycolytic subtype associated with TP53 mutations and high hypoxia-inducible factor (HIF) expression; (4) a fetal liver-like phosphoinositide 3-kinase/mammalian target of rapamycin (PI3K/mTOR)-activated subtype linked to vascular invasion; and (5) a Notch/transforming growth factor-β (NOTCH/TGF-β)-activated scirrhous subtype marked by cytokeratin 19 (CK19) and cholangiocytic marker expression. This classification framework is notable for incorporating the concept of liver zonation—a fundamental aspect of hepatic architecture—into HCC classification. Although the functional compartmentalization of hepatocytes along the portal-to-central vein axis (from periportal to perivenous zones) has long been recognized, few studies have explored its relationship to tumor heterogeneity [
7]. This study demonstrates that, at the transcriptomic level, Wnt/β-catenin activation correlates with perivenous hepatocyte traits (e.g., glutamine synthetase [GLUL] and cytochrome P450 family 3 subfamily A member 4 [CYP3A4] expression), whereas IL6-JAK-STAT3 and RAS signaling are associated with periportal metabolic functions such as albumin synthesis and gluconeogenesis.
A particularly notable finding is the association between glycolysis activation and the loss of liver-specific metabolic functions. Tumors within the glycolytic subtype show increased expression of glucose transporters (glucose transporter type [GLUT] 1, GLUT3, and GLUT5), high uptake on fluorine-18 fluorodeoxyglucose PET (18F-FDG) PET, hypointensity on Gd-EOB-DTPA-enhanced MRI, and histological patterns consistent with compact or macrotrabecular-massive (MTM) subtypes. Conversely, metabolically “rich” tumors are defined by well-differentiated histology, uniform arterial enhancement, elevated Gd-EOB-DTPA uptake, reduced FDG avidity, and suppression of glycolytic gene expression.
Importantly, metabolic reprogramming through glycolysis activation is a well-established adaptation in many solid tumors other than HCC. Notably, enhanced glycolytic activity has been reported in triple-negative breast cancer (TNBC) and ovarian cancer, supporting tumor growth, invasion, and therapeutic resistance. In TNBC, for example, hypoxic cancer-associated fibroblasts (CAFs) secrete colony-stimulating factor 3 (CSF3), activating glycolysis via the CSF3 receptor-phosphoglucomutase 2-like 1 (CSF3R–PGM2L1) pathway and enhancing tumor invasiveness [
8]. In ovarian cancer, the mitochondrial E3 ubiquitin ligase (MARCH5) enhances glycolysis by promoting the degradation of mitochondrial pyruvate carrier 1 (MPC1), further enhancing tumor progression and metastasis [
9]. These findings suggest that the glycolytic switch is not a liver-specific phenomenon, but a common metabolic adaptation observed across multiple tumor types. As such, these highlight glycolysis as a shared metabolic vulnerability across cancer types, supporting its potential as a target for broad-spectrum anticancer therapies.
Aligned with this concept is the authors’ proposal of “physiological dedifferentiation.” Rather than passively losing hepatocellular identity, tumor cells may actively abandon liver-specific metabolic functions in favor of glycolysis-dominant metabolism—thereby facilitating survival and adaptation. This metabolic transition is better understood as a strategic reprogramming that enables immune evasion and metabolic efficiency—enhancing tumor adaptability and resistance. This conceptual shift offers deeper understanding of tumor plasticity and therapeutic failure. From this perspective, cancer cells should no longer be regarded as simple de-differentiated but as adaptive systems continuously evolving survival mechanisms in response to environmental pressures. Future therapeutic strategies—particularly immunotherapies and metabolism-targeting interventions—must account for this dynamic adaptation, extending beyond targeting discrete genetic mutations to encompass functional states and metabolic reprogramming.
Clinically, the proposed classification system has clinical applications across imaging, immunotherapy, and patient prognosis. Its ability to stratify tumors based on non-invasive imaging features is especially valuable in settings patients who cannot undergo biopsy. For example, the IL6–JAK–STAT3 subtype is associated with an inflamed tumor immune microenvironment (TIME) and increased sensitivity to immune checkpoint inhibitors (ICIs), whereas the Wnt/β-catenin subtype shows immune exclusion and poor ICI responsiveness. These findings align with recent efforts to develop transcriptome-based immune signature prediction tools [
10].
In conclusion, this study represents an important step forward by proposing a function-based classification system that overcomes key limitations of earlier gene expression- based models. By integrating spatial physiology, metabolic reprogramming, and developmental origin, this framework enhances our understanding of HCC biology. Ultimately, it may provide a foundation for precision medicine strategies focused on functional vulnerabilities, biomarker-based diagnostics, and individualized therapies.
FOOTNOTES
-
Authors’ contribution
EJJ designed and wrote the manuscript; PSS supervised the whole project and wrote the manuscript.
-
Acknowledgements
This work was supported by the Scientific Research Fund of the Korean Liver Cancer Study Group.
-
Conflicts of Interest
The authors have no conflicts of interest to declare.
Abbreviations
cancer-associated fibroblasts
colony-stimulating factor 3
cytochrome P450 family 3 subfamily A member 4
immune checkpoint inhibitor
interleukin-6–Janus kinase–signal transducer and activator of transcription 3
membrane-associated ring-CH-type finger 5
mitochondrial pyruvate carrier 1
Notch/transforming growth factor beta
programmed death ligand 1
phosphoglucomutase 2-like 1
phosphoinositide 3-kinase/mammalian target of rapamycin
tumor immune microenvironment
REFERENCES
- 1. Yang X, Yang C, Zhang S, Geng H, Zhu AX, Bernards R, et al. Precision treatment in advanced hepatocellular carcinoma. Cancer Cell 2024;42:180-197.
- 2. Suzuki H, Mishra S, Paul S, Hoshida Y. Molecular and immune landscape of hepatocellular carcinoma for therapeutic development. J Liver Cancer 2025;25:9-18.
- 3. Shin H, Yu SJ. A concise review of updated global guidelines for the management of hepatocellular carcinoma: 2017-2024. J Liver Cancer 2025;25:19-30.
- 4. Poté N, Caruso S, Caderaro J, Cauchy F, Lagadec F, Couchy G, et al. Borderline hepatocellular adenomas: A practical diagnostic approach based on pathologic and molecular features. Mod Pathol 2023;36:100211.
- 5. Jeon Y, Kwon SM, Rhee H, Yoo JE, Chung T, Woo HG, et al. Molecular and radiopathologic spectrum between HCC and intrahepatic cholangiocarcinoma. Hepatology 2023;77:92-108.
- 6. Aoki T, Nishida N, Kurebayashi Y, Sakai K, Fujiwara N, Tsurusaki M, et al. Molecular classification of hepatocellular carcinoma based on zoned metabolic feature and oncogenic signaling pathway. Clin Mol Hepatol 2025;31:981-1002.
- 7. Halpern KB, Shenhav R, Matcovitch-Natan O, Toth B, Lemze D, Golan M, et al. Single-cell spatial reconstruction reveals global division of labour in the mammalian liver. Nature 2017;542:352-356.
- 8. Qin W, Chen B, Li X, Zhao W, Wang L, Zhang N, Wang X, et al. Cancer-associated fibroblasts secrete CSF3 to promote TNBC progression via enhancing PGM2L1-dependent glycolysis reprogramming. Cell Death Dis 2025;16:249.
- 9. Xu Y, Zhao S, Shen Y, Li Y, Dang Y, Guo F, et al. MARCH5 promotes aerobic glycolysis to facilitate ovarian cancer progression via ubiquitinating MPC1. Apoptosis 2024;29:1232-1245.
- 10. Yim SY, Lee SH, Baek SW, Sohn B, Jeong YS, Kang SH, et al. Genomic biomarkers to predict response to atezolizumab plus bevacizumab immunotherapy in hepatocellular carcinoma: Insights from the IMbrave150 trial. Clin Mol Hepatol 2024;30:807-823.
Citations
Citations to this article as recorded by
- Correspondence to editorial on “Molecular classification of hepatocellular carcinoma based on zoned metabolic feature and oncogenic signaling pathway”
Tomoko Aoki, Naoshi Nishida, Masatoshi Kudo
Clinical and Molecular Hepatology.2026; 32(1): e79. CrossRef - Reply to correspondence on “Molecular classification of hepatocellular carcinoma based on zoned metabolic feature and oncogenic signaling pathway”
Eun Ji Jang, Pil Soo Sung
Clinical and Molecular Hepatology.2026; 32(1): e115. CrossRef - Fluorine-18 Fluorodeoxyglucose Positron Emission Tomography: A Potential Imaging Biomarker for Predicting Response to Combination Immunotherapy in Hepatocellular Carcinoma
Masatoshi Kudo
Liver Cancer.2025; 14(5): 511. CrossRef
