Correspondence to editorial on “Molecular classification of hepatocellular carcinoma based on zoned metabolic feature and oncogenic signaling pathway”

Dear Editor,

We would like to express our sincere gratitude to Drs. Eun Ji Jang and Pil Soo Sung for their insightful editorial [1] on our recent publication [2]. Our study aimed to propose a novel molecular classification of hepatocellular carcinoma (HCC) by integrating multi-layered data, including metabolic, morphological, and imaging features [3]. We sincerely appreciate your thoughtful editorial, which not only summarized our findings but also provided broader insights beyond the scope of our manuscript. In particular, we greatly appreciate your discussion of tumor immune microenvironments (TIME) and treatment resistance in solid tumors with enhanced glycolysis, and broader clinical implications shared across multiple cancer types.

We would like to supplement your discussion by emphasizing that such metabolic reprogramming toward glycolysis is a phenomenon observed not only in breast and ovarian cancers but also in gastric and lung cancers [4]. As you have noted, targeting the functional vulnerabilities of cancer metabolism holds great promise for personalized medicine. We agree that, particularly in the context of immunotherapy, combining immune checkpoint inhibitors with agents targeting lactate metabolism (e.g., monocarboxylate transporter [MCT] inhibitors) or glycolytic master regulators (e.g., sulfamonomethoxine-derived compounds targeting ALDOA, such as CDP-5), may significantly enhance therapeutic efficacy [5,6].

However, we would also like to offer some important caveats: First, enhanced glycolysis does not necessarily equate to increased lactate accumulation. While our Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis suggested metabolic activity favoring pyruvate and lactate production, our study did not directly quantify lactate levels. It is well known that high lactate concentrations can suppress CD8+ T cell function and promote immunosuppressive tumor microenvironments via upregulation of programmed cell death protein 1 (PD-1) in regulatory T cells (Tregs) and M2 macrophage polarization [4,7]. Thus, future studies should rigorously evaluate the degree of lactate accumulation in glycolysis-high solid tumors. Second, even in tumors with upregulated glycolysis and lactate production, the TIME in HCC is heterogeneous (see Fig. 1). We previously reported that our proposed subclasses show substantial overlap with the Hoshida classification [2]. Among glycolysis-driven HCCs, those corresponding to the Hoshida S1 subtype exhibit relatively high immune scores, while those aligning with Hoshida S2 tend to exhibit low immune scores. This observed heterogeneity suggests the presence of additional, yet unidentified, genetic or epigenetic alterations, which we are actively investigating through ongoing analyses. Notably, UBE2S has been identified as a prognostic gene in HCC, promoting glycolysis through E3 ligase-independent ubiquitination [8]. Moreover, recent studies have implicated the loss of function of activin A receptor type 2A (ACVR2A) as a contributor to poor prognosis in non-viral HCC by facilitating a Treg-enriched TIME via enhanced glycolysis and lactate production [6]. Consequently, increasing attention is being focused on genetic alterations related to glycolysis and lactate metabolism.

Figure 1.

Metabolic subclass distribution and tumor immune microenvironment (TIME) across Hoshida subclasses. Metabolic and signaling subclasses, as defined in our previous study2 are visualized as a heatmap based on Hoshida subclasses: S1 (yellow), S2 (light blue), and S3 (dark blue). Enrichment scores for glycolysis, fatty acid metabolism, and bile acid metabolism were calculated using GSVA with gene sets from the MSigDB Hallmark collection. Immune and stromal scores were estimated using the ESTIMATE, and the proportions of tumor-infiltrating immune cells (CD8⁺ T cells, Tregs, M1 and M2 macrophages) were inferred using quanTIseq algorithm. The expression levels of ACVR2A were normalized and visualized in a heatmap. GSVA, Gene Set Variation Analysis; MSigDB, Molecular Signatures Database; TIME, tumor immune microenvironment; Treg, regulatory T cell; ACVR2A, activin A receptor type 2A.

We would also like to clarify a few potential points of misunderstanding in the editorial regarding our classification system: Our five proposed molecular subtypes are as follows: (1) Wnt/β-catenin-high subtype; (2) Kirsten rat sarcoma viral oncogene homolog (K-RAS)-high subtype; (3) interleukin 6-Janus kinase-signal transducer and activator of transcription 3 (IL6-JAK-STAT3)-high subtype resembling inflammatory Hepatocellular adenoma (HCA); (4) fetal liverlike phosphoinositide 3-kinase/mammalian target of rapamycin (PI3K/mTOR)-activated subtype; (5) Notch/transforming growth factor-β (NOTCH/TGF-β)-activated scirrhous subtype. Subtypes 4 and 5 both exhibit TP53 mutations and high Hypoxia-inducible factor (HIF)-1A expression, and correspond to our glycolysis-driven subclass. Additionally, it may be premature to define the IL6–JAK–STAT3-high subtype as an inflamed type or immune active subtype [9,10]. Although it displays high immune and stromal scores, there is not yet conclusive evidence regarding its responsiveness to immunotherapy. Further validation is warranted.

Currently, we are developing a regression model incorporating dynamic imaging studies, serum biochemical markers, and pathological features to accurately predict glycolysis-driven HCC subclasses. We plan to present this work at upcoming conferences and in future publications. If implemented in clinical practice, this model may allow functional classification of HCC without the need for genomic analyses.

We are truly honored by this opportunity for scholarly exchange. We sincerely hope our research will contribute to the ongoing advancement of Clinical and Molecular Hepatology and the field of liver disease as a whole.

Abbreviations

ACVR2A

activin A receptor type 2A

ALDOA

Aldolase A

CPD-5

Sulfamonomethoxine-derived compound targeting ALDOA

GSVA

Gene Set Variation Analysis

HCA

Hepatocellular adenoma

HCC

Hepatocellular carcinoma

HIF1A

Hypoxia-inducible factor 1-alpha

IL6–JAK–STAT3

Interleukin-6–Janus kinase–Signal transducer and activator of transcription 3 pathway

KEGG

Kyoto Encyclopedia of Genes and Genomes

K-RAS

Kirsten rat sarcoma viral oncogene homolog

MCT1

Monocarboxylate transporter 1

NOTCH/TGF-β

Notch/Transforming growth factor-beta signaling pathway

PD-1

Programmed cell death protein 1

PI3K/mTOR

Phosphatidylinositol 3-kinase/ Mammalian target of rapamycin signaling pathway

TIME

Tumor immune microenvironment

Treg

Regulatory T cell