Dear Editor,
Meticulous radiological classification of Lee et al. [
1] successfully associates the type IV infiltrative morphology with markedly poor outcomes following first-line atezolizumab plus bevacizumab (Ate/Bev) therapy, accompanied by distinct genomic (enriched tumor suppressor protein
p53 [
TP53]/ataxia telangiectasia mutated [
ATM] mutations) and transcriptomic (activated transforming growth factor-beta [TGF-β] signaling, epithelial-mesenchymal transition, immunosuppressive microenvironment) features [
1]. This represents a significant contribution to the field. To further transform these robust associations into more causal and clinically actionable insights, we wish to propose several key considerations that could refine the interpretation of the findings and directly impact their implications for both biology and clinical strategy.
The study compellingly presents the co-occurrence of
TP53/ATM loss-of-function mutations and upregulated pathways like TGF-β as defining molecular characteristics of the infiltrative subtype. However, an intriguing alternative interpretation warrants consideration: these features may not stem solely from tumor-cell-autonomous programs but could also reflect the tumor’s adaptation to, or selection by, a pre-existing fibro-inflammatory liver microenvironment. The hypothesis is particularly relevant for hepatocellular carcinoma (HCC), where chronic liver disease creates a pervasive backdrop of activated hepatic stellate cells, immunosuppressive macrophages, and pro-fibrotic signaling [
2]. Recent spatial transcriptomic evidence suggests that hepatocarcinogenesis may be influenced by the metabolic and oxidative stress gradients within the liver lobule, implying that tumor growth patterns, including infiltrative spread, could be fundamentally shaped by this parenchymal context [
3]. Consequently, the multi-omics signatures identified from bulk tumor tissue may represent an amalgam of intrinsic tumor drivers and extrinsic microenvironmental pressures. Acknowledging this possibility would subtly but importantly refine the conclusion from “these molecular features define the subtype” to “these features are associated with the subtype and may be co-shaped by a pro-fibrotic, immunosuppressive niche.” This distinction is critical as it expands the therapeutic horizon beyond targeting tumor cells alone to potentially combining strategies that modulate the pathogenic stroma.
Furthermore, while the study establishes infiltrative morphology as an independent predictor of poor Ate/Bev response via multivariate analysis, the robustness of this conclusion would be strengthened by accounting for additional key prognostic variables. Two categories warrant consideration. First, tumor mutational burden (TMB), a potentially important genomic covariate whose prognostic role in HCC is still evolving, could serve as a key mediator [
4]. If the
TP53/ATM mutation enrichment correlates with a lower TMB, the primary driver of resistance might be a “cold tumor” phenotype. In this case, infiltrative morphology could be a radiographic correlate of this low-immunogenicity state rather than an independent biological driver. Conversely, a higher TMB might act as a mediating variable. Integrating TMB into the multivariate model would clarify whether morphology retains independent prognostic significance after adjusting for this key genomic metric. Second, the host’s systemic inflammatory and nutritional status, quantifiable by scores like the prognostic nutritional index or neutrophil-to-lymphocyte ratio, are powerful independent prognostic factors [
5,
6]. High tumor burden can exacerbate inflammation and cachexia. If patients with infiltrative morphology have a worse baseline host status, their poor survival may be partly mediated by this ‘host vulnerability,’ potentially overstating the independent predictive value of morphology alone. Future studies should incorporate these host factors, ideally using integrated scores that combine inflammation and nutrition with liver function and tumor burden for superior predictive power [
7], to disentangle their contribution from that of tumor morphology.
Finally, we appreciate the authors’ precise framing of their findings within the context of first-line Ate/Bev. This specificity is, in fact, a major strength of the study. It reveals not a general resistance to immunotherapy, but a precise vulnerability in the “anti-programmed death ligand-1 (PD-L1) plus anti-vascular endothelial growth factor (VEGF)” strategy. This naturally leads to the next, more constructive set of clinical questions: Is the resistance primarily driven by evading VEGF inhibition, PD-L1 blockade, or the synergy of both? The answer would guide subsequent therapy. For instance, dual immune checkpoint blockade (e.g., anti-cytotoxic T lymphocyte antigen 4 [CTLA-4] plus anti-programmed death-1 [PD-1]) has demonstrated efficacy as an alternative first-line regimen [
8]. Importantly, recent pharmacodynamic analyses suggest that CTLA-4/PD-1 dual blockade and PD-L1/VEGF inhibition exert distinct mechanisms of action on the peripheral immune response, supporting their non-cross-resistant potential [
9]. Furthermore, other targeted agents (e.g., against fibroblast growth factor 19 [FGF19]/fibroblast growth factor receptor 4 [FGFR4]) represent a promising direction, as preclinical evidence links this pathway to PD-L1 regulation and immune evasion [
10]. Therefore, the crucial clinical implication of this work is to caution against excluding all other immunotherapies for patients with infiltrative HCC after Ate/Bev failure. Instead, it should motivate the investigation of alternative or additive mechanisms, such as different immune checkpoints or stroma-modulating agents.
In summary, Lee et al. [
1] have identified a high-risk HCC subtype. By integrating analyses of the tumor-microenvironment interface and key biomarkers like TMB, the “infiltrative” classification could evolve from a prognostic phenotype to a more etiologically defined category with clearer therapeutic implications. This progression would ultimately advance patient care from risk stratification towards personalized strategies designed to overcome the specific resistance mechanisms this important study has brought to light.
FOOTNOTES
-
Authors’ contribution
Conceptualization: GD, KY. Methodology: DH. Visualization: GD, DH, KY. Funding acquisition: GD, DH, KY. Project administration: GD, DH, KY. Supervision: GD, KY. Writing - original draft: GD. Writing - review & editing: KY, DH.
-
Conflicts of Interest
The authors have no conflicts to disclose.
Abbreviations
atezolizumab plus bevacizumab
ataxia telangiectasia mutated
cytotoxic T lymphocyte antigen 4
fibroblast growth factor 19
fibroblast growth factor receptor 4
programmed death ligand-1
transforming growth factorbeta
tumor suppressor protein p53
vascular endothelial growth factor
REFERENCES
- 1. Lee WS, Woo S, Lee SH, Jo GH, Kim I, Kim H, et al. Integrative multi-omics profiling identifies infiltrative hepatocellular carcinoma as an immunotherapy-resistant subtype with distinct molecular features. Clin Mol Hepatol 2026;32:258-275.
- 2. Ma L, Li CC, Wang XW. Roles of cellular neighborhoods in hepatocellular carcinoma pathogenesis. Annu Rev Pathol 2025;20:169-192.
- 3. Barrows KM, Porat-Shliom N. Hot zones for liver cancer: metabolic zonation, ferroptosis, and the origins of HCC. Cancer Res 2026;86:1347-1350.
- 4. Gabbia D, De Martin S. Tumor mutational burden for predicting prognosis and therapy outcome of hepatocellular carcinoma. Int J Mol Sci 2023;24:3441.
- 5. Tang HH, Zhang MQ, Zhang ZC, Fan C, Li SS, Chen W, et al. Prognostic nutritional index predicts survival in intermediate and advanced hepatocellular carcinoma treated with hepatic arterial infusion chemotherapy combined with PD-(L)1 inhibitors and molecular targeted therapies. BMC Cancer 2025;25:603.
- 6. Peng J, Chen H, Chen Z, Tan J, Wu F, Li X. Prognostic value of neutrophil-to-lymphocyte ratio in patients with hepatocellular carcinoma receiving curative therapies: a systematic review and meta-analysis. BMC Cancer 2025;25:571.
- 7. Su Y, Liang Y, Zhong D, Yan H, Yang Q, Shang J, et al. Construction and validation of a novel liver function-tumor burdeninflammation-nutrition (LTIN) score for HCC patients underwent hepatectomy. BMC Cancer 2025;25:504.
- 8. Abou-Alfa GK, Lau G, Kudo M, Chan SL, Kelley RK, Furuse J, et al. Tremelimumab plus durvalumab in unresectable hepatocellular carcinoma. NEJM Evid 2022;1:EVIDoa2100070.
- 9. Kelley RK, Lee Y, Conway J, Kurland JF, McCoon P. T-cell receptor and immune gene expression pharmacodynamics for durvalumab alone and with tremelimumab or bevacizumab in unresectable hepatocellular carcinoma. Clin Cancer Res 2026;32:694-704.
- 10. Guo C, Zhou N, Lu Y, Mu M, Li Z, Zhang X, et al. FGF19/FGFR4 signaling contributes to hepatocellular carcinoma survival and immune escape by regulating IGF2BP1-mediated expression of PD-L1. Biomed Pharmacother 2024;170:115955.
Citations
Citations to this article as recorded by
- Correspondence to letter to the editor on “Integrative multi-omics profiling identifies infiltrative hepatocellular carcinoma as an immunotherapy-resistant subtype with distinct molecular features”
Won Suk Lee, Seonjeong Woo, Sohyun Hwang, Chan Kim, Hong Jae Chon
Clinical and Molecular Hepatology.2026; 32(3): e411. CrossRef
