Review Reports - Spatial Transcriptomics Reveals Distinct Architectures but Shared Vulnerabilities in Primary and Metastatic Liver Tumors

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a study of two-specimen high-definition spatial transcriptomics (10x Visium HD) analysis comparing primary hepatocellular carcinoma (HCC) with a liver metastasis. The analyses revealed (i) contrasting spatial architectures, with ordered lineage gradients in HCC versus sharply compartmentalized invasion and plasticity zones in the metastasis; (ii) a “germline/neural-like plasticity hub” in the metastasis; and (iii) a conserved metabolic program of “porphyrin overdrive,” characterized by reduced CYP expression, enhanced electron transport chain components, and upregulation of FLVCR1 and ALOX5, which was also evident in tumor-free livers of tumor-bearing mice. The manuscript has scientific merit, but needs major revisions to address the below:

The main concern is that this is a case study of two specimens, actually 1 specimen from each of the primary HCC and the Liver metastasis. Despite using spatial transcriptomics with high resolution, the small sample size does not allow generalizability, especially that the authors did not perform confirmatory experiments This should be highlighted. I suggest that in the in the abstract the authors change the part related to sample size to “We applied high-definition spatial transcriptomics (10x Visium HD) to fresh-frozen specimens from one hepatocellular carcinoma (HCC) and one liver metastasis patient, profiling over 16,000 genes per sample with >97% mapping rates, as part of a two-specimen exploratory case study.” This also should be emphasized in the discussion.
Absence of experimental validation of the signatures of porphyrin overdrive, FLVCR1, ALOX5. These need to be confirmed with a method other than transcriptomics like IF imaging, IHC, or metabolomics.
Include a pathology/IHC of the lineage/origin of the liver metastasis.
Clarify the Visium HD platform and version. Detail the Visium HD platform features used such as 2 µm features, binning strategy, Software version (Space Ranger v3.0, HD uses version 3.0 not 2.1.1?), filtering thresholds (min UMIs/genes per bin, mitochondrial %), binning method (default 10x HD binning vs custom).etc. Also, include spatial statistics (Moran’s I, Ripley’s K, or neighborhood enrichment), do not rely only on visual inspection to detect zonations and boundaries
For the mouse dataset, results shown without accession number, methods, or statistical rigor. Detail these.
Tables S1 and S3 include some metabolic genes, it is better to have a table summarizing CYP down, ETC up, FLVCR1/ALOX5 expression spatially.
In the discussion add a limitations paragraph. Explicitly state he small sample size (n=2) and lack of generalizability among the other limitations. Lack of experimental validation, etc…
In the abstract, mention that the comparison data set is mouse “against an independent systemic dataset.” mouse dataset. Validation should be in a human dataset. If no human counterpart is available then add to limitations and discuss how this limits the conclusions.
Add the re-analysis of the Vandenbon et al. mouse dataset to the Methods section.
The authors suggest that tumors at distant sites can systematically reprogram distant tumor-free tissues such as the liver. Is this systemic reprogramming tissue-specific? Are all tissues prone to this reprogramming? What tissues are more prone than others?
If the reprogramming such as the “porphyrin overdrive” is not specific to tumor-bearing tissues, then how can the authors claim that the rewiring can offer “potential therapeutic targets that link local tumor niches to systemic host–tumor interactions.” What are the chances of targeted therapy?
Use of human biobank tissue requires IRB information. Include institutional approval number.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This study applies high definition spatial transcriptomics to compare primary hepatocellular carcinoma and liver metastasis, revealing distinct tissue architectures but a shared metabolic rewiring program termed porphyrin overdrive. Despite unique lineage plasticity hubs in metastases, both tumor types converge on altered hemelipid metabolism, suggesting common therapeutic vulnerabilities. The manuscript may be further improved by following suggestions.

The work already used in title should not be repeated in keywords, select correct keywords relevant to the study.
The study is based on only two human specimens (one HCC and one metastasis), which severely limits generalizability; stronger justification for using such a small cohort is needed.
The claim of identifying a conserved porphyrin overdrive program would be more convincing if validated by metabolomics or protein-level assays, rather than relying solely on transcriptomics.
The spatial resolution of Visium HD is high, but it does not achieve true single-cell resolution; the manuscript sometimes overstates single-cell level interpretation without acknowledging this limitation.
While the manuscript highlights distinct spatial architectures between HCC and metastases, it does not adequately address inter-patient heterogeneity, which could confound the observed differences.
The assignment of metastatic tumor origin to colorectal carcinoma is tentative, yet the discussion extrapolates mechanisms as if this were certain; more cautious language or additional validation is warranted.
The proposed therapeutic implications of targeting FLVCR1 and ALOX5 are speculative and lack experimental validation; functional assays or literature-based therapeutic evidence should be incorporated.
Figures (Figures 2-5) contain dense labeling and complex legends; they are difficult to interpret without additional schematic simplification or summary diagrams.
The integration with the mouse dataset is presented as supporting systemic metabolic reprogramming, but the methods for re-analysis and cross-species comparison are not described in sufficient detail.
The manuscript would benefit from including a more comprehensive comparison with existing spatial liver cancer atlases to position the findings in the broader landscape of spatial transcriptomic studies.
The discussion sometimes merges descriptive findings with interpretive speculation; clearer separation between observed data and proposed models would strengthen the manuscript’s rigor.
Provide limitation of the study as a separate heading, may be after discussion part.
Improve conclusion by incorporating global application of the study with future direction.

Comments on the Quality of English Language

May be improved.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In this manuscript, Adapa et.al. profiled the spatial transcriptome similarity and difference between HCC and liver metastasis. Overall, the experiment was well designed, data was well explained and manuscript was well written.

The limitation is that only 2 samples were incorporated in the study (HCC and liver metastasis), where authors also used mouse dataset for assisting validation. I would recommend authors try to find more dataset for cross validation.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The manuscript provides novel and important insights into liver tumor biology and host–tumor interactions. While the small sample size means the conclusions should be considered exploratory, the depth of spatial and molecular resolution makes it a valuable blueprint for future studies. It is suitable for publication in the present form.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 5 Report

Comments and Suggestions for Authors

This manuscript presents a high-resolution spatial transcriptomic comparison of primary hepatocellular carcinoma (HCC) and liver metastasis. The work is timely and relevant, providing novel insights into tumor architecture, lineage dispersion, immune–stromal organization, and metabolic reprogramming. The identification of distinct spatial structures (ordered lineage in HCC vs. invasion/plasticity zones in metastasis) and a shared “porphyrin overdrive” metabolic program is an interesting contribution.

However, the study is based on only two specimens, which raises concerns regarding generalizability. Moreover, some of the interpretations appear overstated relative to the data presented, and several aspects would benefit from functional validation and a more cautious framing.

Major Comments

Limited sample size: The study analyzes only one HCC and one liver metastasis specimen. While the depth of resolution is impressive, the conclusions (e.g., the presence of a germline/neural-like plasticity hub) cannot be considered representative without further validation. The authors should frame these findings explicitly as hypothesis-generating and emphasize the need for validation in larger independent cohorts or publicly available datasets.

Overgeneralization of conclusions: The manuscript concludes that primary and metastatic tumors share “vulnerabilities.” However, these are inferred from two samples only. Phrasing such as “potential shared vulnerability” or “candidate conserved program” would more accurately reflect the exploratory nature of the findings.

Lack of functional validation: The identification of FLVCR1, ALOX5, and oxidative phosphorylation genes as potential targets is compelling, but remains transcriptomic. Protein-level validation, metabolite profiling, or functional assays (e.g., porphyrin accumulation measurement, pharmacological inhibition) would considerably strengthen the biological interpretation. At minimum, closer integration with published proteomic or metabolomic evidence is needed.

The differences in immune clusters (vascular-integrated in HCC vs. matrix-embedded in metastasis) are described mainly in spatial and structural terms. The manuscript lacks deeper immunological analysis (e.g., T cell exhaustion, B cell clonality, cytokine signaling pathways). More nuanced functional interpretation of these immune aggregates would increase impact.

While the identification of metabolic rewiring is important, therapeutic translation (e.g., targeting FLVCR1 or ALOX5) is still highly speculative. These points should be clearly presented as exploratory directions rather than actionable targets at this stage.

Minor Comments

Terminology: Terms such as “germline-like” and “neural mimicry” are somewhat ambiguous. Providing precise marker genes and functional context would improve clarity. Please check the manuscript carefully.

Figures and clarity: Several figures (e.g., Fig. 3 and Fig. 5) are visually dense, making it difficult to follow the key findings. A simplified schematic summarizing core mechanisms would aid comprehension.

Supplementary data usage: Supplementary Tables S1–S4 are under-utilized. More explicit reference and explanation of marker genes and enrichment results in the main text would help readers assess reproducibility.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

My comments have been addressed.

However, The metabolic reprogramming needs better discussion, especially to show that only distant tumor bearing tissues are affected while normal tissues are not.

I think it is a far extension to base the whole metabolic reprogramming notion based on n=2. The conclusions should be treated as preliminary and awaiting larger cohort studies.